{"articles":[{"id":"3f5bf23c-c618-484a-bc77-dceca5c03ff3","tweet_url":"https://x.com/ericzakariasson/status/2096917654179402172","tweet_text":"📝 Grok Bot Marketplace is live!\n\nWe launched the Grok Bot Marketplace last week on Grok Bot · Bot Marketplace! It's a shelf of teammates other people already built for real jobs. Pick one, import it into your sidebar, and you're chatting in one click.\nStart exploring\nTo get going, click Marketplace, then Bots. From here you can start exploring bots from both the Grok Bot team and the community! With a marketplace, you can find bots build specifically for a task you are working on, whethers its engineering, sales or design! Go check out some pre-made ones.\nWe've highlighted a couple of bots in the featured section with some of the most loved ones:\n \nWe've also curated a Grok Bot shelf for the one we actually run at work ourselves. Writing, exec support, customer calls, prospecting, design critique, project ops. Curious to hear what you think of them!\n \nThe most popular Bots right now\nOutside the team, people have been publishing public templates too. Here's a snapshot of the top 10 bots:\ndr eggbot designs a focused bot after a few preference questions.\nResearchy for research with live search and citations.\nlast30days for what people actually said about a topic lately.\nIndex is an SEO/AEO teammate; briefs, not articles.\nPG for prospecting and outreach drafts.\nAlfred designs and governs a company's bot org.\nProjects Manager Manage a team of Bots with a projects board\ntinkabot turns an API into a plugin other bots can use.\nBest Video Editor cuts owner footage after you agree on the treatment.\nProduct Idea Stress Test pressure-tests a founder idea.\nIf you've built a bot that you've found useful, I'd love to see it!\n\nhttps://x.com/i/article/2096917648839954432","summary":"The Grok Bot Marketplace enables builders and founders to discover, deploy, and compose pre-built AI agents for specific workflows—from sales prospecting to design critique—eliminating the need to build specialized chatbots from scratch and accelerating time-to-productivity for common business tasks. For AI practitioners, this represents a shift toward composable agent ecosystems where bots can integrate with APIs and other bots, creating opportunities to build bot-building tools and establish new patterns for how autonomous systems coordinate across organizational functions.","author_name":"eric zakariasson","author_handle":"ericzakariasson","timestamp":"2026-09-07T11:05:02+00:00","comment_count":5,"like_count":30,"retweet_count":4,"submitted_at":"2026-09-07T11:20:03.411559+00:00","created_at":"2026-09-07T11:20:03.411559+00:00","seo_title":"Grok Bot Marketplace launches with community-built bots","bookmark_count":13,"image_url":"https://pbs.twimg.com/media/HRm_e9GbEAAr_gJ.jpg","keywords":["ai assistants","productivity tools","automation","saas","product management","software engineering","remote work","machine learning","entrepreneurship","side hustles"],"video_url":null,"video_thumbnail_url":null,"image_grid_urls":null,"profile_id":"2cf6f0f2-2efd-48b9-ad6a-0ea33110c537","link_card_url":null,"link_card_title":null,"link_card_image_url":null,"link_card_domain":null,"is_pinned":false,"pinned_at":null,"content_type":"article","article_title":"Grok Bot Marketplace is live!","article_image_url":"https://pbs.twimg.com/media/HRm_e9GbEAAr_gJ.jpg","slug":"grok-bot-marketplace-is-live-2096917654179402172","quality_score":5,"noindex":false,"profiles":{"id":"2cf6f0f2-2efd-48b9-ad6a-0ea33110c537","bio":"@cursor_ai & tinkering. http://colf.dev","name":"eric zakariasson","handle":"ericzakariasson","website":"anyblockers.com","location":"San Francisco, CA","verified":true,"avatar_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/avatars/ericzakariasson.jpg","banner_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/banners/ericzakariasson.jpg","joined_date":"2012-02-01T08:00:00+00:00","tweet_count":6783,"listed_count":929,"follower_count":82481,"following_count":601}},{"id":"93441e63-5265-435a-b86a-c9ce3cc16a16","tweet_url":"https://x.com/vasuman/status/2095999742031675738","tweet_text":"📝 Applied AI Doesn't Work\n\nThe world has spent a fortune on AI. For most enterprises, nothing changed. I've spoken to over 300 CEOs, CIOs and CFOs at the largest companies on Earth. I learned most companies apply AI onto their garbage processes, and the end result is just making garbage faster. Whether it's buying thousands of Claude Code licenses, committing $50M in annual token spend, or running AI training sessions over Zoom, I've seen first-hand how inefficient AI adoption has been. It doesn't have to be this way.\nWe've already gone down this path, multiple times in fact, and we ought to learn from history when it comes to AI adoption. My hope is that by the end of this article, AI leaders around the world stop with their archaic and naive methodology of AI application, and instead follow the tried and true process illustrated below.\nPart 1: History and its Parallels\nIn 1990, a former MIT computer science professor by the name of Michael Hammer wrote an article for the Harvard Business Review. He had already spent years inside companies that spent enormous amounts of money on computers yet achieved very little ROI, and he was among the first to uncover why.\n\"Heavy investments in information technology have delivered disappointing results, largely because companies tend to use technology to mechanize old ways of doing business. They leave the existing processes intact and use computers simply to speed them up.\"\nThat's a real quote. Are you kidding me? If you swapped out 'information technology' for 'AI', you could publish that tomorrow and be on the front page of Business Insider. In the same piece he wrote that it is time to stop paving the cow paths. It's been 36 years since, and guess what we're still doing?\nNote: To pave the cow paths means to automate, formalize, or lock in an existing inefficient process instead of redesigning it for how things should work. Saved you the trip to Claude, you're welcome.\nMeanwhile, over $1T was spent in capex, building out AI (Microsoft, Alphabet, Amazon, and Meta alone spent $410B in 2025; eye-watering, I know), yet only $37B was spent by enterprises on actually using it. What that means in practice: The UK's Department for Business and Trade ran a Microsoft 365 Copilot trial and published the results. 1,000 licenses were rolled out over 3 months, and there was only an average of 1.14 Copilot actions per user per day. PowerPoint slides came out 7 minutes faster (18 minutes down to 11), but at half the quality score. Excel analysis was actually completed more slowly (and also worse quality). Email time savings were, and I quote, \"extremely small.\" Their own conclusion, verbatim: \"We did not find robust evidence to suggest that time savings are leading to improved productivity.\" And yet 72% of users said they were satisfied or very satisfied.\n \nLong story short: the people liked it, yet they were producing exactly what they were before. Ethan Mollick (Wharton Associate Professor and AI Researcher) put it best: \"AI use that boosts individual performance does not naturally translate to improving organizational performance.\" This article will expand on why that is, and offer solutions that we've seen work for us in practice.\nFor context, I'm Vas, the CEO of @varickagents. We work with extremely large companies (between $500M and $100B in revenue, 1k to 100k employees), helping them adopt AI, from general strategy, to process re-engineering, to building, deploying, and managing their agents. This article is based on learnings we've derived: what separates AI pilots that fail (no measurable ROI) from those that succeed (80%+ of the work is handled entirely by an agent, saving time and money).\nThe reason this keeps happening isn't sheer stupidity. Some of it is that, of course, but most of it is inertia and design. Whoever signs an AI contract today may own tooling and vendor selection, but they certainly do not own the process. Nobody in that chain is empowered to walk into finance, for example, and say that their 14 step process should actually only be 5 steps. So nobody does it. Instead, you 'apply AI' on what you currently have, and you end up with faster sh*t.\nAI isn't supposed to be 'applied'. That doesn't work. You 'apply' a coat of paint. Too many people are thinking about AI like a fresh coat of paint, and paint doesn't fix what's underneath. Nobody ever said 'applied cloud' or 'applied digital'. No, instead they said Digital Transformation, Cloud Modernization, because everyone understood that those were transformations, and that the technology was only a part of them. AI takes the same shape.\nBack to Hammer: his best number is from an insurance company he studied. An application took 22 days to move through the business. Of those 22 days, the total time anybody spent working on it was 17 minutes. Suppose, today, your AI guy drops the best model in the world onto that process and it makes every individual step twice as fast. He's now saved 8 and a half minutes. He chalks that up as 50% time savings, gets his promotion and bonus, and rides off into the sunset. It's 8 minutes.\n \nThe savings were not in working on the application. They were in the queues and handoffs, in the waiting periods for other teams, and the second round of reviews. Spending your energy speeding up the 17 minutes is worse than doing nothing; you just spent your budget, and proved to your board that 'AI doesn't work'. You might think this is a relic of the 90s, but it's not. We worked with a company that went through this exact problem, from its own system logs. Opening up a new client case took 25 minutes of real work. The elapsed time, from the case being entered to it going active, ran anywhere from 2 days to 2 weeks. The work can be minutes while the process can still be weeks.\nWhen I say most AI programs are making sh*t faster I mean that literally. The process is sh*t. AI is making it faster. To which I've heard: \"well we're a massive company, so clearly we're doing something right, and we don't need to change.\" Adapt or die is alive and well today, more so than ever, and this is the one place AI differs from digital and cloud. Used properly, AI gives a competitor leverage over you that those never did. Whether it's to undercut you on cost, scale infinitely without processing headcount, or outreach the entire world before you can put together a lead list, they will crush you. Best case you continue growing, and now your headcount scales exponentially. 3 analysts become 6 but the 6 now need a manager, and given it's 2 teams, they need a handoff that didn't exist before, and now that handoff needs a status meeting and a PM, etc etc. This is why headcount charts bend upwards, while revenue-per-employee doesn't. Adapt or die, but don't pretend AI isn't happening. Your competitors are stopping at nothing to figure this out.\n \nPart 2: What to do Today\nYour data is a mess and your systems don't talk to each other, so maybe your first instinct is to fix this before touching anything else. You just need one ERP, one CRM, one data platform, then AI on top. That sounds great in theory, but IT specialists understand that it is never this simple. Take a look at what this actually costs, in both money and time.\nTSB spent £318M on a core banking migration, ate roughly £200M more in incident costs, was fined £48.65M by the FCA and PRA in December 2022, and broke banking for 5.2 million customers. Zimmer Biomet, an $8B manufacturer, filed a $172M claim against Deloitte in September 2025, after a botched 'migration and consolidation' left them unable to perform basic functions, like ship their product, invoice their customers, or produce functional sales reporting. And those are just the two examples that made the news; incidents like these are far more common than you may think.\nWhat you should focus on settling is process, instead. Say you've bought 9 companies over the last 20 years, meaning you now have the same process that runs 9 different ways across your 9 subsidiaries. If you want to 'apply' AI across your company, you'll have to build the same agent 9 different ways, and then be forced to maintain all 9 sets of agents, in perpetuity. To avoid this purgatory, you need to first define the single process that runs globally, which you roll out across your entire entity, before you even think of applying AI on top of that.\nConversely, you don't NEED one ERP. Deterministic integrations used to break the moment that 2 systems disagreed about what a vendor record was. However, agents tolerate this inconsistency, so an orchestration layer that sits across the systems you already have IS the solution to disparate systems. Your systems don't just need to talk to each other, they need to be happily married, tolerating each other's faults.\nStart mapping\nTo redesign your business you need to first understand it. You don't get to this level of understanding by picking a single workflow, where you spend 8 weeks perfecting something that turns out to be gated by a sister process upstream, and end up with net 0 ROI.\nAt the same time, however, you can't just ingest the entire business at once, you simply must start somewhere. Start at the department level. It's not going to be perfect - finance touches sales every time revenue recognition depends on how something was sold. HR touches IT and Ops on every new hire. But at least this way, you identify 'bundles' of workflows that feed into the same department (for example, Finance contains AP, AR, FP&A, Billing, Banking, Reconciliations, etc), and you can begin mapping out the following 7 details:\nThe 'happy path': what is this workflow responsible for, and what happens in the most ideal scenario. For most companies, this is about the only thing that's truly documented to a T.\nThe exceptions. This is where most of your time goes. Don't just describe what happens, go in and calculate what share of volume leaves the happy path, where it goes, who gets pulled in, what the cycle time is for these exceptions to be handled, and what the cost of an error is here.\nWhat's upstream and downstream of this workflow? How does this process interact with those, when things are late or go wrong in each?\nWhat are all of the systems of record involved, and which one wins when 2 of them disagree?\nHow does this differ by region, by entity, by subsidiary, by which acquisition it came in through?\nWhat is the touch time vs the elapsed time, and how does the gap vary at every single step?\nWhat is every single person in the function's AI-ability, and what degree of ownership should they have over a future-state AI system? (Calculating this is subjective and varies heavily).\nThere are two procedures required to extract the above information:\nProcess mining against the systems of record to farm actions, timestamps, throughput, edits, and the reality that does not match documentation.\nInterviews, because the above will give you data but conversations with operators will tell you exactly where the pain is. The reality of your organization lives in the heads of 10-20 people who have been doing this exact work for your exact company for years, and at the end of the day your AI needs to be adopted by these people in order for it to work effectively.\nDoing only one of these steps leaves you with an incomplete picture, and is the biggest reason why AI discovery fails. I've seen Applied AI firms try to take shortcuts with 'AI Interviews', or by sending engineers to do the interviews. This doesn't work. Half the job is developing a strong relationship with the operators, and knowing what questions to ask them, so that you leave with a good understanding of what to build and how to deploy it.\nAgents that work aren't just LLM calls\nProcess redesign is a sorting exercise that requires software and AI experience. Every step falls into one of 3 buckets:\nDeterministic. If X then Y, with no judgment involved. For example, if the invoice is under a certain dollar amount threshold and it matches the PO & goods receipt, then you pay it. If the disclosure names a competitor, then route it to the conflicts team. These actions are best as plain, good old software. You don't need a fancy LLM call if you can do it instead with regular code, which is cheap, auditable, deterministic (never hallucinates), and fast.\nAgentic. If you have thousands of past examples of humans making a judgment call with the outcome recorded, and this step is low enough risk (for example, deal desk approvals, or GL coding), you should absolutely use an LLM for judgment. Especially if the output is a yes/no, a route-to-this-team, a flag/don't flag, or a match/don't match, ship this with an agent and measure every output. On the other hand, if the output required is 5 pages of a document that a human has to read in full to verify anyways, then you're better off hybridizing the approach. Here, you need to judge the difficulty of the action, not the difficulty of the thinking.\nHuman-in-the-Loop. For steps that are too risky, or don't have enough historical context, you're better off keeping the human in the loop, because the cost of an agent getting an action wrong is far too high. Instead, put every relevant piece of evidence in front of that human at the moment they decide. For example, an agent will detect a mismatch between an invoice and a purchase order, and then present the human with 3 options: approve anyways, reject plainly, or provide feedback and route to someone else. The agent just saved your human 40 minutes of hunting through emails, invoice records, and more, so that instead, they can make a decision in 30 seconds. But the final action is still gated by that human.\nOnce you've sorted every step into one of the 3 buckets above, you can turn something like a 25 step process into a cluster of 3 agents, with deterministic steps before and after, and finally 2 points of human-in-the-loop action to unblock. This process re-engineering work, in conjunction with the agents themselves, is where all of the ROI lives.\n \nAfter this, you baseline KPIs before building anything. Your goal is to calculate the numbers behind the process you're solving for, including where and how those numbers are measured today. If you can't do that, this project is destined to fail. You need to be able to look back 12 months from now and say: we used to be at X, and now we are at 3X, or 10X, or 100X, thanks to the agents that we built. Only then can you say this project was truly a success.\nFinally, you prioritize and sequence your build. First, tackle the workflow with the most handoffs, as this is likely the lowest hanging fruit where you can see the clearest results. The key distinction is that this is not necessarily the same workflow with the most volume.\nWhat does this look like in practice?\nWe re-engineered a global business reconciling 300+ bank accounts every month. 2/3 of them were handled by an offshore team, and the rest split across 2 other regions in 3 different formats, but entirely in Excel. There were 12,000+ open reconciliation items, and 4 days at the start of every month were spent obtaining statements. Their reconciliation process took almost the entire month, every month.\nTheir workflow before we re-engineered was as follows (and before you make fun of them, recognize you definitely have processes like this, if not worse, at your own company):\nChase and collect statements from each bank, in whatever format they provide\nNormalize them into the regional workbook\nPull the ledger extract\nMatch line by line\nFlag what doesn't match\nEmail a local controller for context on the flagged items\nWait for their reply\nChase the reply if unanswered\nFinally get the reply, then post the adjustment\nRoll the region into the consolidated file\nDo that entire 1-10 step process again for each of the 2 other regions, each in its own format.\nClose.\nIf you make each of those steps faster, the matching gets faster. The issue is, matching was never the problem. The 4 days of waiting for statements, and the email to the controller in another timezone, and the reply still taking 3 days, are. By making matching faster, you would shave hours off a process that is measured in weeks.\nWe instead redesigned their process to be 3 steps.\nStatements arrive on a feed, conveyor belt style, instead of being chased. This has absolutely nothing to do with AI.\nEverything that can be matched based on rules will be matched automatically, and if not, an agent will assemble the case end-to-end: with the item itself, historical categorizations, evidence gathered from various systems of record, and a proposed next step.\nFinally, a human works through their queue of exceptions, with the evidence prepared for their review, instead of them having to manually sift through data scattered across ERPs and spreadsheets.\nChasing, emailing, waiting, re-keying, and regional roll-ups were eliminated entirely. This is the difference between an agent system that saves minutes and hours, and an agent system that saves weeks.\n \nPart 3: Why isn't this more common?\nYou were lied to\nThe AI pitch you received was misleading from the start. AI adopted as a point solution is not something that you derive value from. You cannot do this from just IT, and you cannot do this from an 'innovation function' with a $1M budget and 0 authority. Redesigning a process that spans 4 departments requires somebody who can tell all 4 departments that the process is changing. There are only a handful of people in a company that can really do this; it has to come from the top.\nBut another degree of complexity: the people actually doing the work today, e.g. the head of accounts payable, and the operators and analysts across finance, also need to be included. They have to want the new AI system, because they have to adopt it and push it tomorrow. They have to want to divulge information that they think can potentially replace them, even if that is not the case a majority of the time.\nAnd most importantly, adopting AI is seen as risky. The status quo is 'safe' for most people, who get no promotion if AI works, yet get fired if it doesn't, even if sticking to the status quo means letting the company die a slow death over time.\nRe-engineering is quite hard\nHammer and Champy's own estimate in 1993 was that as many as 50% to 70% of the organizations that attempted process re-engineering did not achieve the results they intended. (Clearly not much has changed since the 90s). Hammer himself said later that he had been \"insufficiently appreciative of the human dimension.\" What he meant was that 'removing a handoff', at that time, meant firing the person responsible for that handoff (that was their entire job), and then asking everyone else to change their work, while the 'software' of the time didn't support such a change. Which means that after the consultants came and went, the org structure and process quickly went back to its original state: one riddled with inefficiencies and errors.\nToday, however, two things are fundamentally different. First, AI-native tooling offers an unprecedented fluid software state: one that can adapt to the changes around it. An agent sitting in your system of record can run handoffs that used to rely on people, simultaneously handling discrepancies around it from deterministic software, and is also immune to variance common in humans. Second, the work that you are eliminating first is almost entirely coordination, that is: chasing, waiting, escalating, triaging, etc. After these process improvements, and a real AI transformation, all that remains is high-judgment, high-leverage work, which your best people should be spending all of their time doing anyways.\nA brave new world\nTraditional consultancies are decent at process work, I'll give them that. Indeed, if you want an elaborate illustration of a 15-step process that spans 10 countries and 3 entities, your run-of-the-mill consulting firm will produce the most beautiful slide you have ever seen, for $300,000. What this consulting firm cannot do is the latter, more important half, which is knowing what a model can and cannot be trusted with, what an agent that runs 40,000 times in a month costs, which model (Fable 5.1, vs Gemini 3.8 Flash) to use for which style of task (GL Coding vs Deal Desk Routing), how to design an eval suite optimally for a particular process, and which of the 3 buckets above (Deterministic, Agentic, HITL) a given step in a process truly belongs in. And thus, if you go this route, your redesign is created by consultants who are guessing, as opposed to experienced AI-native strategists and FDEs, and 6 months later you wonder why your pilot crash-landed.\nAI labs and companies, on the flip side, have the opposite problem. They can build an agent (most of the time, at least), but they've never sat down with both finance and operations to break down a workflow where both departments disagree about ownership and who is upstream of whom, so instead they'll slap AI onto a broken process and call it automated, instead of gutting and redesigning that process like it should have been months ago. Furthermore, the biggest 'applied AI' labs like OpenAI (DeployCo) and Anthropic (Ode) will force you to marry their models for life. Betting your entire company's intelligence stack on a single provider is suicide. Tomorrow, as they rate limit, quantize, price-hike, and retire the model you're running on, you're left with the mess, and the bill. You may think it's as simple as swapping an endpoint from Claude to GLM or Gemini or DeepSeek, but it isn't. Rerunning evals and redetermining which level of thinking to use per model call is a painstaking process, and the labs will not help you do this.\nWhat you need instead is the best of both worlds combined: people who have actually run a finance, operations, sales, or HR function, combined with the engineering talent to ship production-grade software into ERPs like Microsoft Dynamics or CRMs like Salesforce, combined with AI engineers who can build model-agnostic company-harnesses with built-in enterprise-level governance and eval-suites. Some refer to these people as 'Forward Deployed Engineers', but the truth is most FDEs today are mediocre engineers, and mediocre consultants, with absolutely no AI ability, and you end up with basically nothing being done well.\n \nWhat does this look like when you do it right?\nThere are 4 levers that matter when you're transforming a company with AI: cost (ideally reduced), revenue (ideally increased), time (ideally faster), and risk (ideally lower). Across our deployments, roughly half of them at public companies, here are examples of what actually moved, lever by lever:\nTime. We brought the month-end close process down from 18-22 days to 7-9 days. Supply chain disruptions are now triaged in under 6 hours, as opposed to several days.\nRevenue. Deals used to sit idle for an average of 23 days; now the average is 6 days, and rep admin time dropped from 38% of the week to 14%. This time went straight back into selling.\nCost. AP exception handling went from 600-800 invoices per month to under 50. In a marketing org, 8-12% of the media budget was recovered from pacing errors that used to surface only at month end, saving millions per quarter.\nRisk. Payroll corrections fell 90%, because the reconciliation across systems now ran before payroll instead of after. 80% of employee support queries were answered from the company's own policy documents, while edge cases still escalated to a human in HR.\nIn total, the value delivered across this set of deployments was over $100M. The best part is that this number was co-created with our clients; it's not just us saying it, it was measured.\nWhat does a real AI Transformation Company look like?\nEverything described above is how we do things at Varick, and it's the only approach that has worked for the companies we've rolled AI out for. These are incredibly large enterprises, with thousands of people, dozens of systems of record, and infinite complexities, yet the playbook is the same, and it works every time.\nOur pod of Strategists and Engineers come in at the department level to first map how the work actually happens, not just accept how it's documented, through a combination of process mining and interviews, calculation of elapsed time vs touch time, mapping of the exceptions, upstream effects, and downstream effects, and understanding of who owns what. Instead of months, we complete this process in just a few weeks, at which point you receive your redesigned process, the value to be captured, custom baselined KPIs, and the priority ranking order in which the agents would get built. Next, we would begin building the agents, which come equipped with enterprise-level governance, auditability, eval-suites, and monitoring out-of-the-box.\nWe work with companies doing over $500M in revenue, tackling their toughest workflows, across product, finance, sales, procurement, operations, HR, and more. If you want to see what this looks like against one of your own processes, book a call on our website, and come prepared to talk about your least favorite workflows, and what makes them so complex. And if you found this content helpful, subscribe to our newsletter, where we put out weekly write-ups on the latest in 'Applied' AI, and everything we've learned by doing.\nTLDR\nMost of the AI spend in the enterprise today is being applied on top of the exact same processes that were already there, which means all you are really doing is making a bad process run faster. Michael Hammer said this about IT back in 1990, and the UK government basically proved it again in 2025 with 1,000 Copilot licenses: 1.14 actions per user per day, no productivity gain that they could find, and 72% of the users were satisfied anyways.\nThe time was never in the work itself, it was in the queues and the handoffs. Hammer's insurance application sat for 22 days in process, with 17 minutes of actual work in it. Even if you make the work twice as fast, you have saved 8 minutes.\nConsolidating your systems first (one ERP, one CRM, one data platform) is a well documented way to lose 2 years and 9 figures before you see a single dollar of value. What you actually need is one definition of the process and one owner, because agents can tolerate systems that disagree with each other, and deterministic integrations never could.\nStart at the department level, mine the systems of record and interview the operators, and then sort every single step into one of the 3 buckets: deterministic software, an agent, or a human in the loop with the evidence prepared for them. Done properly, a 25 step process becomes 3 agents with 2 human checkpoints.\nWhen we did it this way, month-end close went from 18-22 days down to 7-9, AP exceptions went from 600-800 a month down to under 50, payroll corrections fell 90%, and the total across our deployments came out to over $100M, measured together with the clients.\nNone of this works unless it comes from the top, the operators actually want it, and whoever you hire has process people, software engineers, and AI engineers all in the same room, and almost nobody has all three.\n\nhttps://x.com/i/article/2095378321588826112","summary":"# Why This Matters\n\nMost enterprises are wasting billions on AI by automating broken processes instead of redesigning them first—a lesson from 1990's IT transformation failures that remains unlearned, as evidenced by the UK government's Copilot trial showing 1.14 actions per user per day with no productivity gains. For builders and founders, this reveals that sustainable AI competitive advantage comes not from better models but from rearchitecting workflows to eliminate handoffs and queues, requiring cross-functional ownership and the rare combination of process expertise, software engineering, and AI capability working together.","author_name":"vas","author_handle":"vasuman","timestamp":"2026-09-04T22:17:34+00:00","comment_count":14,"like_count":102,"retweet_count":10,"submitted_at":"2026-09-04T22:20:03.76571+00:00","created_at":"2026-09-04T22:20:03.76571+00:00","seo_title":"Most Enterprise AI Projects Waste Money on Broken Processes","bookmark_count":205,"image_url":"https://pbs.twimg.com/media/HRZ5qYabkAADtzK.jpg","keywords":["process redesign","business process optimization","enterprise automation","ai adoption strategy","workflow automation","operational efficiency","digital transformation"],"video_url":null,"video_thumbnail_url":null,"image_grid_urls":null,"profile_id":"ddd0250f-d31a-4ebe-b9f2-a0ae18778f70","link_card_url":null,"link_card_title":null,"link_card_image_url":null,"link_card_domain":null,"is_pinned":false,"pinned_at":null,"content_type":"article","article_title":"Applied AI Doesn't Work","article_image_url":"https://pbs.twimg.com/media/HRZ5qYabkAADtzK.jpg","slug":"applied-ai-doesn-t-work-2095999742031675738","quality_score":8,"noindex":false,"profiles":{"id":"ddd0250f-d31a-4ebe-b9f2-a0ae18778f70","bio":"CEO @varickai | AI Agents for Enterprise: http://varickagents.com | prev. @meta","name":"vas","handle":"vasuman","website":null,"location":"manhattan","verified":true,"avatar_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/avatars/vasuman.jpg","banner_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/banners/vasuman.jpg","joined_date":"2020-06-01T07:00:00+00:00","tweet_count":6905,"listed_count":1124,"follower_count":91362,"following_count":465}},{"id":"047f8366-a30b-4645-ae6b-eb35b149ee68","tweet_url":"https://x.com/AndrewYNg/status/2095890279865721217","tweet_text":"📝 AI Engineering Skills Map: Using coding agents\n\nA key AI engineering skill is using coding agents. Your skill at steering them both to write code and to carry out non-code tasks, such as analyzing data or managing system operations, allows you to get a lot more done.\nThe rapid pace of evolution for coding agents means this skill, too, is evolving rapidly — faster than other top-level AI engineering skills. Proprietary agents (like Claude Code, Codex, and Cursor) and open agents (like OpenCode and Pi) progress in strides via both harness and model improvements. So keeping up with how to use coding agents requires a continuous process of experimentation, building, and learning.\nIn interviewing dozens of top AI Engineers and reflecting on our own team’s use of coding agents, we found a consistent high-level workflow for building software with them. The key steps are:\nPlanning. This includes (i) brainstorming, which may include research, experimentation, and understanding the existing codebase (if any) and (ii) writing a spec that captures requirements, technical design, and architecture, followed by generating an execution plan. You might also review the plan to interrogate key assumptions and check for security, overengineering, and other gaps.\nExecution, where you build, test, and verify, with the right balance between agent autonomy and human oversight. This involves (i) having the agent build the software, with a calibrated level of agent autonomy and (ii) verifying its output via automated and/or human checks.\nDeployment and monitoring, in which you (i) deploy, perhaps gated with a CI/CD pipeline or additional human gates, and (ii) use agents to watch logs, surface issues, and propose and execute improvements.\nThis high-level workflow is similar to the one typically used to build software before coding agents. Now, we focus much less on code and instead focus on deciding what to build, designing the architecture, writing the spec, and verifying outputs.\nThe duration of each step can vary significantly between projects, and steps can be omitted. For example, the spec for a greenfield (meaning built-from-scratch) prototype might be loosely described in a quickly written prompt, whereas the spec for a brownfield (pre-existing) project with many users might require much more effort to write and verify. Further, the workflow is highly iterative, and skilled developers know when feedback from a later step should lead them back to an earlier one. For example, if verification fails, they know how to steer the agent to rebuild and fix errors; or if monitoring surfaces issues, how to have agents update the system and redeploy.\nTo use coding agents effectively in this workflow, the key skills are:\nDirecting the workflow\nEnabling agent autonomy\nReviewing the work\nCustomizing the agent and its environment\nCoding agent foundations\nDirecting the workflow. You know how to navigate each step of the workflow above. This involves deciding how much human and how much agent effort to spend on each and when to go back to an earlier step to iterate. It requires deeply understanding the tradeoffs of speed, cost, technical risk, and human effort, so you can decide how much to research and plan up front, when to retain human ownership over critical work, how to choose the architecture, how much detail to write into a set of planning artifacts (like a spec), and how to decompose the work into verifiable steps.\nEnabling agent autonomy. When applying a coding agent to the steps in the workflow, you choose the autonomy level: Do you watch it and go back-and-forth interactively or delegate a larger chunk of work to it? And when do you set a clear goal and have it loop until it succeeds? Additionally, you have to manage the context carefully for the agent. As the build proceeds through different phases, you will calibrate when to make sure key learnings, user feedback, and assumptions — including assumptions that changed partway through the build — are captured for the agent to use downstream. Additionally, you will decide when to set up many agents to run in parallel on a decomposition of the task — either by having a human or a higher-level agent orchestrate these other agents — and how to manage human attention across concurrent agent sessions. You also know how to run agents safely, setting permissions and gating actions appropriately to let development proceed quickly while limiting the risk of leaks, data loss, or other damage.\nReviewing the work. The output of a coding agent is uncertain. We don’t know in advance what good ideas it might come up with and what bugs it will implement. Reviewing and verifying the output is a key step to ensure you are getting the result you want and to redirect the agent if not. You will design testing and validation that is matched to the task, applying both behavioral and functional verification as needed. You might also test user flows, perhaps having an agent provide screenshots as evidence of success or failure. For qualitative/behavioral evaluation, eval sets, perhaps with LLM-as-a-judge, can be used.\nYou also need to decide how much of these tests should be automated. Some workflows will have all testing and validation fully automated so the agent can check its work and know when it has succeeded in completing a task. You have to evaluate the tests to ensure they correspond to your aims, and you will evolve them if not. Additionally, you use agentic code review and run AI-enabled security and architecture audits. When AI review isn’t sufficient, you judiciously insert human reviews of the code behavior (and, infrequently, of code as well) while exploring how to automate this review further. Finally, you verify deployment and can operationalize monitoring and incident management with agents.\nCustomizing the agent and its environment. Your ability to update both the agent and the environment it works in allows your agents to efficiently get the context they need, access tools, and build correctly and efficiently. You know how to integrate agent skills, plugins, and MCP servers. Occasionally you will prune them when they are no longer necessary (such as when a new model obviates an old skill). You can use hooks to automate repeatable parts of the development process, like triggering automated code reviews or CI/CD pipelines. You can also maintain the environment the agent works in: updating the standing context (such as AGENTS.md or CLAUDE.md) with information on the codebase, key architectural assumptions, code style, and data access patterns. You know how to preserve state across multiple sessions and across parallel agents, and accumulate agent learnings over time, perhaps by running post-run retrospectives to capture what did and did not work. You also know how to set up consistent conventions and structure to make your codebase navigable to the agent, and how to occasionally clear out agent-generated debt. When you work in a team, you consider how to coordinate context across different developers’ agents.\nCoding agent foundations. Finally, to make good decisions throughout, you have a good understanding of how coding agents work: how they carry out codebase search/retrieval, how they manage their context windows, how different operations (like adding tool calls, MCP servers, etc.) affect context, how agents and subagents interact, and how the agent is built by wrapping a harness around an LLM. This makes the agent less of a black box and helps you to recognize failure modes, such as overengineering a simple solution, losing rigor because the agent lacks an explicit verification process, stopping short of the goal, or agent actions that risk destruction of files or production data. It also helps you reason about the agent’s state and steer it by giving it the right prescription or context. And when monitoring a run, this understanding allows you to better spot when the agent goes off-track and requires your intervention.\nI find that social media often gives oversimplified descriptions of how to use coding agents. For example, it is sometimes useful to get agents to run autonomously for hours and burn millions or tens of millions of tokens. But currently the practical utility of very long-horizon tasks — especially relative to their cost — has been amplified beyond reality. Instead, most effective coding agent use is a complex, highly iterative process, and being able to intervene with high-skill judgement gives much better results.\nYour skill at using coding agents will make you an effective builder. This positions you to also steer the overall build. I will say more about this in next week’s letter.\n\nhttps://x.com/i/article/2095882148670832640","summary":"Mastering coding agents—particularly knowing when to direct agent autonomy, review outputs, customize environments, and understand their technical foundations—is now a critical AI engineering skill that evolves faster than traditional software skills and directly determines builder productivity. This skill matters because effective coding agent use is a disciplined, iterative workflow requiring judgment calls about planning depth, verification rigor, and human oversight rather than simply delegating tasks to agents autonomously, enabling builders to ship significantly more capability per unit of effort.","author_name":"Andrew Ng","author_handle":"andrewyng","timestamp":"2026-09-04T15:02:37+00:00","comment_count":112,"like_count":3762,"retweet_count":536,"submitted_at":"2026-09-04T15:35:03.986052+00:00","created_at":"2026-09-04T15:35:03.986052+00:00","seo_title":"Effective Coding Agent Use Requires Iterative Human Oversight","bookmark_count":5936,"image_url":"https://pbs.twimg.com/media/HRYTehda0AAcCvW.jpg","keywords":["ai engineering","coding agents","software engineering","machine learning","artificial intelligence","automation","programming"],"video_url":null,"video_thumbnail_url":null,"image_grid_urls":null,"profile_id":"3ecf579e-e33f-4b35-8ac3-36de3de7ad8e","link_card_url":null,"link_card_title":null,"link_card_image_url":null,"link_card_domain":null,"is_pinned":false,"pinned_at":null,"content_type":"article","article_title":"AI Engineering Skills Map: Using coding agents","article_image_url":"https://pbs.twimg.com/media/HRYTehda0AAcCvW.jpg","slug":"ai-engineering-skills-map-using-coding-agents-2095890279865721217","quality_score":8,"noindex":false,"profiles":{"id":"3ecf579e-e33f-4b35-8ac3-36de3de7ad8e","bio":"Co-Founder of Coursera; Stanford CS adjunct faculty. Former head of Baidu AI Group/Google Brain. #ai #machinelearning, #deeplearning #MOOCs","name":"Andrew Ng","handle":"AndrewYNg","website":"andrewng.org","location":"Palo Alto, CA","verified":true,"avatar_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/avatars/andrewyng.jpg","banner_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/banners/andrewyng.jpg","joined_date":"2010-11-01T07:00:00+00:00","tweet_count":2020,"listed_count":20240,"follower_count":1803269,"following_count":1104}},{"id":"cbecaa72-7d87-4722-b865-aa204a8499f3","tweet_url":"https://x.com/github/status/2095248954133016876","tweet_text":"📝 August 2026 Ship Log\n\n💬 GitHub Copilot is now in Slack and Microsoft Teams\nWorking with your team in Slack or Microsoft Teams? Copilot can join you.\nThis new GitHub integration brings the agentic capabilities of GitHub Copilot CLI and the GitHub Copilot app right to your conversations. Mention GitHub to plan changes, investigate problems, and relay coding tasks.\n \n🤖 New models in GitHub Copilot (and limited-time discounts 👀)\nHave you tried these new models yet?\nGemini 3.7 Flash, MAI-Code-1.1-Flash, and Kimi K3 (hosted by @FireworksAI_HQ) are all generally available in any surface you use GitHub Copilot.\n \nPlus, special pricing: GPT-5.6 Sol has been reduced by 50% while Auto model selection is now 30% off for Copilot Max users. Try them out in Copilot before the promos end. ⏳\n \n \n🎧 New episodes of the GitHub Podcast\nGet to know your hosts for Season 2. Find it wherever you listen to your podcasts.\n \n💪 Copilot code review now has Balanced depth\nYou now have more choices in effort level for Copilot code review.\n \n➡️ Follow us for more updates, news, and events from GitHub this month, including GitHub Copilot Day, coming September 10. Hear directly from the people building Copilot and the developers putting it to work. Set a reminder. 🔔\n\n\nhttps://x.com/i/article/2095240317981315072","summary":"GitHub Copilot's expansion into Slack and Teams, combined with new model options (Gemini 3.7 Flash, MAI-Code-1.1-Flash, Kimi K3) and configurable code review depth, enables builders and AI practitioners to integrate agentic coding assistance directly into existing team workflows without context-switching. The limited-time pricing on GPT-5.6 Sol (50% off) and Auto model selection (30% off for Max users) provides cost-effective opportunities for founders to evaluate and adopt multiple frontier models for production use before promotional periods end.","author_name":"GitHub","author_handle":"github","timestamp":"2026-09-02T20:34:13+00:00","comment_count":7,"like_count":32,"retweet_count":3,"submitted_at":"2026-09-02T21:20:03.682936+00:00","created_at":"2026-09-02T21:20:03.682936+00:00","seo_title":"GitHub Copilot integrates with Slack and Microsoft Teams","bookmark_count":2,"image_url":"https://pbs.twimg.com/media/HRPMG6QagAAQn1d.jpg","keywords":["ai coding assistant","github copilot","software engineering","machine learning models","code review automation","programming tools","developer productivity","artificial intelligence","saas tools","remote work collaboration"],"video_url":null,"video_thumbnail_url":null,"image_grid_urls":null,"profile_id":"835d6cc6-dd91-48b2-95ac-d5dfa1bcdada","link_card_url":null,"link_card_title":null,"link_card_image_url":null,"link_card_domain":null,"is_pinned":false,"pinned_at":null,"content_type":"article","article_title":"August 2026 Ship Log","article_image_url":"https://pbs.twimg.com/media/HRPMG6QagAAQn1d.jpg","slug":"august-2026-ship-log-2095248954133016876","quality_score":6,"noindex":false,"profiles":{"id":"835d6cc6-dd91-48b2-95ac-d5dfa1bcdada","bio":null,"name":"GitHub","handle":"github","website":null,"location":null,"verified":false,"avatar_url":null,"banner_url":null,"joined_date":null,"tweet_count":3,"listed_count":0,"follower_count":0,"following_count":0}},{"id":"6c72ee9f-23c4-4613-8ba9-f3635267f187","tweet_url":"https://x.com/AmirMushich/status/2095182776249049456","tweet_text":"📝 How to design a Brand System from 1 reference image\n\nFor 10+ years, I've been creating branding & ads for Warner Music, PepsiCo, and others.\nNow I combine that experience with AI to build hi-end visual systems faster. \nThis article shows my complete branding workflow with AI\nThis time, I turned one reference into an identity, brand guidelines, packaging, key visuals, and a scalable campaign world.\nAt the end, I’m also sharing the open-source Lovart Brand System Skill - a copy-paste workflow that guides Lovart Agent through the same process.\nThis article is sponsored by @lovart_ai. Thanks to the team for supporting this post and my work for you.\n \n \nONE REFERENCE -> BRAND SYSTEM\nI wanted to test one question:\nCan an AI agent understand why an image works -> not copy it -> and turn that logic into a new brand?\nSo I found a Pinterest image I like, sent it to the Lovart Agent and gave it a minimal brief of the brand I want to build:\nBrand: Juicee\nProduct: natural, caffeine-free energy drink\nAudience: active people aged 20–35\nFlavors: watermelon, melon, apricot, raspberry\nFrom that, we built a visual identity, a complete multi-slide brand system, four connected SKUs, campaign scenes, packaging concepts, videos & more.\nLovart accelerated the design process - and my role was to judge & direct the outputs, remove conflicting directions, define the rules & system.\nMy workflow:\nReference → Deconstruction → Anchor (Brand Kit) → Guidelines → Brand Lock → Campaign Assets → Packaging\n \nTurn one reference into a brand system with Lovart Agent: → amirmushich.link/design_agent\n \n1. DECODE THE REFERENCE\nI found the reference of the Pickle Potion drink - and this visual gave us four transferable rules:\nSplit-screen composition\nReal ingredient photography\nBold typography\nSimple starburst elements\nand more (see below).\n \nPickle Potion uses those rules to communicate briny shock.\nFor my brand - Juicee - I translated them into a different idea:\nThe intensity of freshness - real fruit that hits.\nThe instruction I used\n \nImportant: A reference should inform the logic - not the design output.\n \n2. FROM DRAFT TO BRAND ANCHOR\nThe first Brand Kit looked okay - but mixed too many styles, patterns, effects, and typography directions.\nIt was content, not a system.\n \nMy feedback was simple:\n“Use more white space, less graphic devices, and one clear direction.”\nMy first job was not to generate more. It was to remove.\nThe problem was a lack of hierarchy:\nRealistic liquid materials, cartoon mascots, icons, product graphics, and several micro-directions were all competing inside one board. \nDifferent materials did not match the graphic language, so the brand felt assembled from fragments instead of directed as one world.\nI rejected it - because it was creatively inconsistent.\nCorrect the direction (if needed):\n \nThe second version became a great anchor for the future brand - one approved image the entire brand could obey.\nHere's how the whole process looks like:\n \nOn this stage, we locked:\nlogo concept;\ncolor palette;\ntypography (free Google Fonts);\nbrand graphics (vector starbursts);\ncan design;\nmedia rule - \"real fruit photography\";\ndesign foundation.\n \nThe system is now locked - let's scale it.\n \n3. TURN THE ANCHOR INTO A BRAND SYSTEM\nFrom one anchor, Lovart Agent created a connected set of brand-guideline modules:\nLogo usage · Typography · Color · Graphic elements · Key visual rules · Packaging · Brand usage rules · Design rules · Photo guidelines\n \nThe same rules then produced 4 flavor key visuals with one repeatable mechanism:\n[fruit on the left] → [product on the right] → [one headline] → [one starburst claim] → [one consistent hierarchy]\nThe SKUs feel different without becoming four different brands.\n \n \n4. EXPAND THE CAMPAIGN (and keep the brand consistent)\nOnce the design system was locked, I introduced 9 additional scene references - to expand the brand campaigns world.\nLovart Agent quickly translated those references into a broad range of explorations - from fruit top views and product close-ups to lifestyle scenes. \nBecause each reference carried its own strong visual language, some early outputs began to over-weight its fonts, colors, patterns, and packaging cues alongside the composition.\n \nThis wasn’t a lack of generation capability - it was a creative-direction challenge. The Agent needed a clearer hierarchy for deciding which source controlled each part of the output.\nOne of Lovart’s strengths is that you can combine references and steer the system iteratively. I used that flexibility to define a clearer reference hierarchy - a workflow I call Brand Lock.\nBrand Lock isn’t a button or a single prompt. \nIt’s a methodology for deciding what each reference is/isn’t allowed to control.\nThe final scene must still be materially original - not a recreated artwork with a different product.\n \nBrand Locking with Lovart Agent\ncombine the scene references, Brand Kit, and packaging board → generate new branded scenes → identify and correct visual drift → reveal the complete scaled canvas\n \n \nThe references could now influence the scene language without owning the scene or redesigning the brand.\nBy repeating the same Agent workflow across different references and flavors, I expanded one approved system into dozens of connected visuals.\nAgent generated the assets\nBrand Lock kept the system consistent.\n \nPrecision editing after generation\nThe editing tools came after generation. Text Edit, Touch Edit, and Edit Elements were useful for precise corrections to assets.\nRefine one completed asset with Text Edit → Touch Edit → Edit Elements\n \nThe same connected system also produced front, three-quarter, isometric, and exploded packaging views.\nNote: These are conceptual packaging visualizations - not final dielines, vector artwork, prepress files, production-ready 3D models, or manufacturing specifications.\n \n \n5. TURN THE BRAND SYSTEM INTO MOTION WITH SEEDANCE 2.5\nOnce the visual system was locked, I used Seedance 2.5 inside Lovart to generate the motion assets, then assembled and finished the final Juicee ad in Adobe.\nThe same approved packaging, colors, typography, and campaign language became the source of truth for video - so motion extended the system.\nThe Agent kept the creative workflow connected - from static brand assets to video generation on the same Canvas. The brand/character consistency is really strong here.\n \n \nONE REFERENCE TO A COMPLETE BRAND WORLD\nThe clearest proof of the workflow is my Lovart canvas.\nReference → Brand System → Entire brand world.\n \nBuild a consistent brand system with Lovart Agent\n→  amirmushich.link/design_agent\n \nMY BRANDING SKILL (GitHub)\nI packaged the complete workflow as a copy-paste Skill for Lovart Agent\nInstead of running prompts manually, paste the Skill into a new Lovart project and let the Agent guide you through:\nmass-user-friendly onboarding;\nreference deconstruction;\nAnchor Brand Kit development;\ncreative-direction critique;\nfonts selection;\ngeneration approval gates;\nbrand guidelines;\nBrand Lock;\ncampaign assets;\nconsistency audits;\npackaging concepts.\nIf you don’t know how to answer his questions - or why a visual direction isn’t working - the Skill can help diagnose the problem before you spend more credits.\nGet the Brand System Skill on GitHub\nhttps://amirmushich.link/brand_skill\n \n\n \n5 THINGS TO REMEMBER\nOne solid anchor beats twenty disconnected outputs\nUse references for principles (not for copy-paste)\nApprove the concept before spending credits on assets\nUse the Agent to generate complete assets - and editing tools to refine the details\nAI accelerates concepting; final production still needs human vision\n \nTHE REAL LESSON\nWe started with one Pinterest image.\nWorking with Lovart, I turned it into an identity, SKUs, brand guidelines, campaign scenes, packaging concepts, and a complete brand world.\nThe agent accelerated exploration and execution. My job was to choose the direction, reject visual contradictions, and define what could change - and what had to stay locked.\nAI can create endless images. A brand begins when someone decides what must stay in its visual language.\nTurn your next reference into a brand system with Lovart Agent\n→ amirmushich.link/design_agent\nOne reference can become more than an image. \nIt can become an entire brand world.\n#BuiltWithLovart #AiDesign #AIBranding\n\n\n\n\n\n\nhttps://x.com/i/article/2094805443335831552","summary":"# Why This Matters\n\nThis post demonstrates a practical methodology for scaling design systems using AI agents, showing that builders and founders can now compress weeks of branding work into iterative cycles by using AI to enforce consistency rules rather than generate infinite variations. For AI practitioners, it reveals the critical insight that AI agents excel at applying design logic and maintaining brand coherence when given explicit constraints (Brand Lock methodology), but still require human judgment to establish hierarchy, reject visual contradictions, and make directional decisions that distinguish cohesive brands from assembled fragments.","author_name":"AmirMušić","author_handle":"amirmushich","timestamp":"2026-09-02T16:11:15+00:00","comment_count":18,"like_count":204,"retweet_count":9,"submitted_at":"2026-09-02T16:20:02.97054+00:00","created_at":"2026-09-02T16:20:02.97054+00:00","seo_title":"Building a Complete Brand System from One Reference Image","bookmark_count":413,"image_url":"https://pbs.twimg.com/media/HRI-fGJW4AAbCgC.jpg","keywords":["visual design","brand identity","design systems","ai design","branding workflow","product design","design consistency","packaging design","brand guidelines","creative direction"],"video_url":null,"video_thumbnail_url":null,"image_grid_urls":null,"profile_id":"39cc71d3-cd77-4c71-b23f-1a2a22606964","link_card_url":null,"link_card_title":null,"link_card_image_url":null,"link_card_domain":null,"is_pinned":false,"pinned_at":null,"content_type":"article","article_title":"How to design a Brand System from 1 reference image","article_image_url":"https://pbs.twimg.com/media/HRI-fGJW4AAbCgC.jpg","slug":"how-to-design-a-brand-system-from-1-reference-image-2095182776249049456","quality_score":6,"noindex":false,"profiles":{"id":"39cc71d3-cd77-4c71-b23f-1a2a22606964","bio":null,"name":"AmirMušić","handle":"AmirMushich","website":null,"location":null,"verified":false,"avatar_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/avatars/amirmushich.jpg","banner_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/banners/amirmushich.jpg","joined_date":null,"tweet_count":69460,"listed_count":1060,"follower_count":74860,"following_count":2112}},{"id":"9ce2d437-2411-4385-8f26-0e1d0845aa02","tweet_url":"https://x.com/jacksondahl/status/2095151581507289540","tweet_text":"📝 21 Lessons from Dialectic with Thrive's Miles Grimshaw\n\nAlways Another Adventure: Miles Grimshaw on Software’s Future, Enduring Companies, and Being Impatiently Patient\nMiles is drawn to beginnings: the new country, the unfamiliar market, the next mountain. I talked to him about how he finds and partners with exceptional people, which genetics shape enduring companies, what moats will survive in software, and beginning again, over and over.\nThe reward for reaching one summit is another climb. A coach sent Miles Camus’ classic essay imagining Sisyphus happy. He’s internalized it with regard to his infinite game as a venture investor: “I love going back down and looking at new summits and going, ‘Can we go do that one?’”\nCourtship starts before the round exists. Miles flew to Ottawa after one call with Turbopuffer’s CEO Simon, offered him a term sheet, got turned down, and spent more than a year helping out anyway. When the call finally came at 9:30 on a Monday night, Thrive was committed by 10:30. “Our challenge is to be impatiently patient.”\nUnderwrite the person before the model. Miles invested in Cursor after getting rough numbers over dinner without ever receiving detailed revenue numbers. “It never starts with a spreadsheet. It starts with that person, the future that they see.”\nInvesting is enlisting. Miles treats early investment like joining the team and working for that ownership over the next decade. After Cursor’s Series A, Thrive went all-in on recruiting for them through the Series B. The relationship is working when a founder thinks, “Sure, I want to raise money from these people, but actually, I kind of want them on the team.”\nInstinct comes from exposure to the best. Your read on great is only as good as the greatness you've seen up close. When Thrive met Mesh Optical (a team who built Starlink's laser interconnect at SpaceX), they went from first meeting on Friday afternoon to commitment by Monday, all in a domain they \"had no business knowing anything about.\"\nSpecific optimism is the job. He once told a candidate, “There are infinite ways this can go wrong that I can enumerate. Our job is to figure out the path of it going right.”\nWhat is it? And who cares? Two questions underpin Miles’s investment analysis. You can say many words about a company, but if you can't simply define what the customer is actually buying and who values it enough to pay, nothing else matters.\nVision is where the founder is looking. Patrick Collison was looking beyond payments toward the GDP of the internet; Cursor’s Michael Truell was looking beyond the IDE toward new abstractions for coding. “In driving and biking, you go where you look… I think that is really true for companies, too.”\nAsk what a company’s growth is made of. Investors love to use a company’s ARR growth as a scorecard: how fast to $10M or $100M? Miles is less interested: “That's a phenotypic expression of an underlying set of genetics.” A handful of large contracts or a long tail of small customers might both have strong ARR, but will grow into completely different organisms.\nPick the customer that helps you live in the future. For early-stage companies, he cares more about demanding, innovative customers than pure revenue. They force the product and the organization to adapt to a changing market. \"I think Stripe would be a very different company if Delta was its first customer instead of Shopify or Lyft.\"\nA great product can outrun its sales team. When he began working with Cursor, it got to tens of millions in ARR with zero salespeople. The sales team remained very small until early 2026. Michael relied on his team’s feedback and individual user conversations because “they had a very strong purity to delivering a great product.”\nCopy the cause of success, not its privileges. Google’s core business was so powerful that 20% time could look like a management principle instead of something Google could uniquely afford. “In some sense, it was the wrong thing to learn the lesson of from that company.”\nSoftware’s hidden moat was wall-clock time. Switching costs mattered, but Miles thinks the undercounted advantage of software was the years spent building a place where customers, integrations, expertise, and mindshare naturally gathered. “The great software companies basically built a city... If you're Benioff and you're building a CRM, sure, you're building a CRM, but you're kind of building the City of Sales.”\nGet big or go luxury. AI has compressed the build time that protected incumbents in software, threatening the middle. Miles suggests a lesson from ecommerce: either cover enormous surface area or be nimble enough to build bespoke, white-glove products.\nCraft creates trust. Customers may not consciously notice every product decision, especially as software becomes more like services. But they feel the accumulated attention to detail and infer that the invisible details have been handled with equal care.\n“I don’t know” is only half the answer. The follow-up is \"I'll figure it out.\" Learning requires enough humility to say “Sorry, I just don’t understand that” and enough self-confidence to believe you can reach the ground truth.\nInstinct must become legible before it becomes useful. Repetition improves Miles’s intuition, but an investment team cannot act on a feeling trapped inside his head. “If I can’t communicate it to the people who trust me the most… how is the rest of the world going to do that?”\nRaise the ceiling in good times and absorb the shock in bad ones. A lesson from Sequoia’s Sir Michael Moritz. Success gives a company the capacity for the exceptional recruit, ambitious customer, or defining move: “when things are working, you need to set your eyes further out.” In adversity, the founder has already felt every problem more deeply than the board will, so constructive support is more useful than drilling on known problems.\nHold jam sessions, not jury trials. He rejects the investment committee as a jury trial where one person presents, and everyone else looks for guilt. “The best version of investment team conversations… is like being in the music studio making music together.”\nStable foundations make larger adventures possible. Miles spent much of his life associating the next adventure with leaving: England, the US, India, China, a new major at the end of college, then Thrive for Benchmark. Returning taught him something else: “I’ve come to find a lot more of the magic that gets unlocked by a strong foundation.”\nEndurance grows from evidence, not encouragement. Miles ran his first marathon at seventeen because he wanted to do it before turning eighteen; years later, he and his wife completed a 150-mile race in New Zealand that half the field did not finish. Belief follows proof: “Once you know you can, you can.”\n \nOther platforms:\nTranscript & links\nSpotify\nApple\nSubscribe on Substack\nYouTube\n\nhttps://x.com/i/article/2095147529155321856","summary":"This conversation distills practical wisdom about identifying exceptional founders, building lasting business relationships, and recognizing the structural shifts AI is creating in software—insights that directly shape how venture investors evaluate opportunities and how founders should think about competitive advantage in a world where build times are compressing. For builders and AI practitioners specifically, the emphasis on \"get big or go luxury\" as software's middle gets squeezed, combined with lessons on product-market fit rooted in customer selection rather than sales velocity, offers a framework for navigating the current landscape where AI tools can rapidly commoditize certain capabilities.","author_name":"Jackson Dahl","author_handle":"jacksondahl","timestamp":"2026-09-02T14:07:17+00:00","comment_count":0,"like_count":40,"retweet_count":2,"submitted_at":"2026-09-02T14:50:06.576485+00:00","created_at":"2026-09-02T14:50:06.576485+00:00","seo_title":"For builders and AI practitioners specifically","bookmark_count":32,"image_url":"https://pbs.twimg.com/media/HRN1tddXYAA2zaq.jpg","keywords":["venture capital","founder mindset","product strategy","software moats","startup investing","customer experience","growth strategy"],"video_url":null,"video_thumbnail_url":null,"image_grid_urls":null,"profile_id":"3ccb2a93-b58e-4d81-ae08-14ddfd5171de","link_card_url":null,"link_card_title":null,"link_card_image_url":null,"link_card_domain":null,"is_pinned":false,"pinned_at":null,"content_type":"article","article_title":"21 Lessons from Dialectic with Thrive's Miles Grimshaw","article_image_url":"https://pbs.twimg.com/media/HRN1tddXYAA2zaq.jpg","slug":"21-lessons-from-dialectic-with-thrive-s-miles-grimshaw-2095151581507289540","quality_score":6,"noindex":false,"profiles":{"id":"3ccb2a93-b58e-4d81-ae08-14ddfd5171de","bio":"@dialecticpod","name":"Jackson Dahl","handle":"jacksondahl","website":"dialectic.fm","location":"NYC","verified":true,"avatar_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/avatars/jacksondahl.jpg","banner_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/banners/jacksondahl.jpg","joined_date":"2008-08-01T07:00:00+00:00","tweet_count":23579,"listed_count":680,"follower_count":41844,"following_count":3687}},{"id":"5d0e1c85-6b0d-4bd3-b845-09b4c7e9c380","tweet_url":"https://x.com/thisisgrantlee/status/2095149908345201048","tweet_text":"📝 You’re trapped in a doom loop. Here’s how to get out. \n\nFor some, it sounds like the “permanent underclass,” or your hard-earned degree becoming obsolete. For others, it sounds like the flatline of all things creative.\nIf this feels familiar, you may be stuck in the AI doom loop. Don’t worry: escape is possible.\nThe doom loop takes different forms: social media threads predicting an imminent mass extinction, perpetual references to Terminator or Ex Machina, and generalized feelings of nihilism.\nLike most fears, it’s self-propelling. It can even be a tantalizing rabbit hole to fall into, where each AI news clip serves as a guaranteed confirmation that the End Times have arrived.\nThis is not to merely wave off or dismiss AI-related fear. Today’s technological advancement is steeped in uncertainty, and uncertainty leads us to fill in the gaps. It is important to acknowledge, however, how these gaps are steered by negativity bias.\nNegative information carries more psychological weight than positive information that is equally as intense. When pessimistic hypotheses flood the zone of most online discourse surrounding AI, the doom loop is fed.\nNotably, the doom loop is fed by the same fears that have paired with most technological advancements to date.\nEven more notably, technology rarely delivers on our fears in the form we predict.\nThe fear of technology cannibalizing creativity, for instance, is nothing new. In 1906, composer John Philip Sousa led a campaign against recorded music, believing it would “ruin the artistic development of music in this country.” This campaign made its way to a congressional hearing, lobbied alongside artistic powerhouses Mark Twain and Victor Herbert.\nHe believed it would eradicate live music. It would retire the use of our vocal cords. It would eliminate the “amateur” musician. Music teachers would become “without field or calling.”\nDo these concerns sound familiar? More importantly, do they sound fulfilled?\nWhile Sousa argued that phonographs would make music soulless, recording technology later played a crucial role in the rise and spread of R&B music…\n…which gave birth to the subgenre we literally call “Soul.”\nRecording music created a new sector of job opportunities, a new industry, and a new mode of creative expression. AI is actively doing the same.\nFeeding this form of fear until it becomes repetitive and ruinous is not the act of preservation it is framed as. More often than not, the doom loop becomes an exercise of imaginative negativity.\nFor Sousa, it was complete creative destruction, something we directly see pervading discourse today. It was that children would become mere extensions of technology, or, in his words, “human phonographs.”\nMy argument is that experience trumps speculation. That’s the way out.\nFear of the dark bred monsters under the bed. It is only the people who investigate with light that find there are no monsters.\nIn the same vein, most of the intense reactions to AI are coming from the sidelines, and people who have sworn off AI cannot see how it can amplify the very characteristics they fear losing. To escape the doom loop, I urge you to see AI as an opportunity.\nExperiment. Play around. Research the ways in which AI can enhance your interests, your creative sparks, and your role at work.\nMaybe it’s placing one “shower thought” into AI each day and seeing where it goes.\nMaybe it’s using a platform that handles the tedious tasks in your workload so you can spend more time focused on creative work, becoming the employee whose ideas are irreplaceable.\nMaybe it’s testing out a new medium for those ideas and seeing what inspiration erupts.\nToday, techno-optimism is often discarded as one of two things: naivete or willful neglect. For me, it is embracing innovation, if not for my own individual productivity, then for my sanity.\n\nhttps://x.com/i/article/2090915316415283200","summary":"The post matters because it reframes AI anxiety as a pattern repeating throughout technological history—fears that rarely materialize in predicted forms—and argues that builders and practitioners escape paralyzing doom loops by engaging directly with AI tools rather than remaining on the sidelines. This shift from speculation to hands-on experimentation is critical: historical precedent (like recorded music creating rather than destroying the music industry) suggests AI will likely generate new job categories, creative possibilities, and industries rather than deliver the apocalyptic scenarios dominating discourse.","author_name":"Grant Lee","author_handle":"thisisgrantlee","timestamp":"2026-09-02T14:00:38+00:00","comment_count":8,"like_count":42,"retweet_count":5,"submitted_at":"2026-09-02T14:30:03.000759+00:00","created_at":"2026-09-02T14:30:03.000759+00:00","seo_title":"Escaping the AI doom loop through hands-on experimentation","bookmark_count":52,"image_url":"https://pbs.twimg.com/media/HQRsbvRawAA52kD.jpg","keywords":["artificial intelligence","psychology","creativity","innovation","technology adoption","leadership","future of work","mental health","productivity","social media"],"video_url":null,"video_thumbnail_url":null,"image_grid_urls":null,"profile_id":"f67551b5-6a43-4e2b-ab18-632e4c62fbfb","link_card_url":null,"link_card_title":null,"link_card_image_url":null,"link_card_domain":null,"is_pinned":false,"pinned_at":null,"content_type":"article","article_title":"You’re trapped in a doom loop. Here’s how to get out. ","article_image_url":"https://pbs.twimg.com/media/HQRsbvRawAA52kD.jpg","slug":"you-re-trapped-in-a-doom-loop-here-s-how-to-get-out-2095149908345201048","quality_score":4,"noindex":false,"profiles":{"id":"f67551b5-6a43-4e2b-ab18-632e4c62fbfb","bio":"Co-founder of @GammaApp","name":"Grant Lee","handle":"thisisgrantlee","website":"gamma.app","location":"Startup →","verified":true,"avatar_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/avatars/thisisgrantlee.jpg","banner_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/banners/thisisgrantlee.jpg","joined_date":"2020-10-01T07:00:00+00:00","tweet_count":2848,"listed_count":343,"follower_count":42928,"following_count":1138}},{"id":"fd0984ec-55fb-479f-baf9-fa89cec53093","tweet_url":"https://x.com/mattyp/status/2094833468400447618","tweet_text":"📝 Using Grok Bot: 8 templates to get inspired\n\nTemplates are now available for Grok Bot, so I wanted to share my favorites.\nTemplates contain the skills, memories, and official @bot plugins from your bots. When you use a template, you get a copy of the bot without any private information.\nTemplates are like sharing a recipe: you're making the same dish, but in your kitchen with your ingredients.\nThat makes templates the perfect way to share workflows without worrying about sensitive data.\n \nI've found templates to be a great way to share the bots I use every day. Today I'll walk through a few of my favorites, from avoiding parking tickets in SF to how I write code at @SpaceXAI.\nHere are my favorite bots, plus templates to get you started.\nTech demos (DevRel bot)\nThis bot helps me try out new technologies.\n \nEach morning, it looks at my @X bookmarks, picks one new tool, and sends over a draft prompt. Each prompt uses a template I derived from my Loops bot (more on that soon).\nI gave the bot some explicit verification parameters: screenshots and videos. I can quickly inspect the deployments from within @Bot.\nThese are actually coming from inside the Cursor Agent where the code is being written, not Grok Bot. The Grok Bot can talk to the Cursor Agent and use its computer, in addition to its own.\n \nThe benefit of using a single repo with Cloudflare is that I can get a preview deployment per project. I have my Cloudflare account hooked up so I can see everything.\nThat means I can:\nSee a quick preview live in @Bot\nOpen the app and play with it in my browser\nDrill into the code or continue iterating in @cursor_ai\nI don't usually end up publishing the demos, but I've found I have a better understanding of what exists, and I'm more likely to use or recommend the libraries in my projects.\nPlugins & connections: @X, @GitHub, @Cursor_ai, @Cloudflare\nTemplate: https://x.ai/bot/zvkkoHMbclsUBWX8MRGpU\nLoops (swe)\nLoops is my software engineering bot.\nI talk to Loops, and Loops talks to a coding agent (my Cursor agent). Loops is my \"outer loop\" engineering agent.\nSince @Bot is automating most common tasks, we can start to think at the meta level. For most engineering, I'm sending through similar prompts and skills, so we can encode those in a bot.\n \nLoops helps me understand background, context, and patterns. Then I kick off agents using my preferred prompting style: pstack, with heavy /goal and loops usage.\nAs an example, say I'm making an update to our website. I might ask Loops to look at contributions from @leerob, @ericzakariasson, and @Jaaneek to understand patterns.\nI might ask it to look at our competitors' documentation to understand how they handle the problems. I might ask how we can verify our solution.\nLoops provides a draft prompt and kicks off cloud agents. Then my workflow shifts from @Bot to @cursor_ai.\n \nPlugins & connections: @GitHub, @Cursor_ai\nTemplate: https://x.ai/bot/Ub3T7usX-c6yRQibQq83P\nX algorithm\nShould I post this now or wait?\nThis bot uses the X Algorithm and our native X connection to advise me on when to post, how to format tweets, and how to engage on the platform.\n \nA disclaimer: I've found that good content always wins over small optimizations, but this bot is super useful if I'm second-guessing myself on how to plan out my content.\n \nPlugins & connections: @X\nTemplate: https://x.ai/bot/X_P19IvPAHZ3FiA1Q-05s\nWriting bot\nThis is my bot for prose editing and writing.\nI rooted this bot in my favorite books on writing: Style: Lessons in Clarity and Grace, On Writing Well, and a few others. It turns out reading is still useful.\nWith Typefully + Notion connectors, it revises my prewritten drafts. I write by hand first, then Writing Bot helps me with grammar and tricky sentences.\nSometimes, I use Writing Bot to pre-populate the details of a piece, like this one.\n \nIt went through all my bots and asked them for a summary, then dropped a draft in Typefully for me to start writing.\nPlugins & connections: @Typefully, @NotionHQ\nTemplate: https://x.ai/bot/gJ4waNMuQoJkQCGX77yF3\nPersonal CRM\nThis bot looks through my X mutuals. It categorizes them cleanly in a Notion board I can sort by country, state, and city.\nNow when I'm traveling, I know exactly who to hit up for a coffee or a walk (yes, I'm the annoying coffee guy).\nThe hard parts of a personal CRM were always information matching and maintenance. This bot handles both of those for me.\n \nFor more details, check out my latest video.\n \nPlugins & connections: @X, @NotionHQ\nTemplate: https://x.ai/bot/FtxHtWPnLGheNJmF04wSf\nImage generation\nI'm starting to experiment more with content creation in Grok Bot.\nMy Image Generation Bot uses a set of tools to generate standardized icon and character sets.\nHere's how it works:\nThe bot has a skill to send standardized prompts to Grok Imagine.\nThose are piped to a @fal background removal model, then I apply a standardized palette.\nThis workflow lets me generate stylized icons and assets that feel thematically consistent.\n \nYou can encode media creation workflows inside Grok Bot to standardize assets for your projects.\nPlugins & connections: @SpaceXAI, @fal\nTemplate: https://x.ai/bot/phPQtGzCZOynABubl0pwx\nParking in San Francisco\nI'm not really sure why, but I named this bot Skippy. It helps me avoid parking tickets in SF.\n \nI can also save spots and get a weekday morning ping before a window hits. It's pretty simple: it just uses open public data for San Francisco.\nI think the next step would be giving it all of those parking apps and having it pay for parking for me.\nI like this bot because it's a clean example of a useful workflow that's as simple as checking a database or website. We can shift that work to a bot that lives on our phone.\nTemplate: https://x.ai/bot/b00VrDzeasoqtPymoNDB-\nDial Bot\nThis bot can make phone calls for me.\nA phone is a tool - you might give an employee a phone and a phone number so they can help you with tasks. With a bit of setup, you can give your Grok Bot a phone.\n \nWith @usebland, I cloned my voice and configured their MCP and calling skills. Now my bot can navigate phone trees, handle holds for me, and more.\nPlugins & connections: @usebland\nTemplate: https://x.ai/bot/NJXi2SWEuhNxjOjspMMPi\nCreating templates and sharing plugins\nTo create a template, open your bot settings (right panel) and select \"Share as template.\"\n \nYour bot will package things up for you.\n \nHitting \"publish\" gives a live link.\n \nTemplates can be shared at the team level or globally.\nTemplates are the first step to more \"social\" bots that you can use with your teammates and share with folks online.\nIn some sense, templates are the first true way to share knowledge work. We've never before had a way to encode the skills, memories, and connections in a workflow.\nGrok Bot templates let you do just that. There's already been an amazing community response to templates. I can't wait to see what our creative users will build.\n\nhttps://x.com/i/article/2094833461630808064","summary":"Grok Bot templates enable builders and founders to share production workflows—including encoded skills, memories, and integrations—without exposing sensitive data, creating a new distribution model for AI-powered knowledge work similar to open-source software. For AI practitioners, this represents the first practical infrastructure for capturing and replicating complex multi-agent systems (like the \"Loops\" engineering bot that orchestrates between planning and coding agents), making it possible to leverage battle-tested automation patterns across teams and communities.","author_name":"matt palmer","author_handle":"mattyp","timestamp":"2026-09-01T17:03:13+00:00","comment_count":21,"like_count":588,"retweet_count":45,"submitted_at":"2026-09-01T17:15:34.626401+00:00","created_at":"2026-09-01T17:15:34.626401+00:00","seo_title":"Eight Grok Bot Templates for Workflow Automation","bookmark_count":1142,"image_url":"https://pbs.twimg.com/media/HRJXw7vbsAEFC7J.jpg","keywords":["ai automation","software engineering","workflow automation","programming","machine learning agents","productivity","content creation","writing","social media marketing","product design"],"video_url":null,"video_thumbnail_url":null,"image_grid_urls":null,"profile_id":"c1863c0d-71d0-4f8a-841b-777cf4a421ae","link_card_url":null,"link_card_title":null,"link_card_image_url":null,"link_card_domain":null,"is_pinned":false,"pinned_at":null,"content_type":"article","article_title":"Using Grok Bot: 8 templates to get inspired","article_image_url":"https://pbs.twimg.com/media/HRJXw7vbsAEFC7J.jpg","slug":"using-grok-bot-8-templates-to-get-inspired-2094833468400447618","quality_score":7,"noindex":false,"profiles":{"id":"c1863c0d-71d0-4f8a-841b-777cf4a421ae","bio":"developer relations, keeping the reps lit @replit","name":"matt palmer","handle":"mattyp","website":"palmer.gg","location":"san francisco","verified":true,"avatar_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/avatars/mattyp.jpg","banner_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/banners/mattyp.jpg","joined_date":"2022-04-01T07:00:00+00:00","tweet_count":9081,"listed_count":502,"follower_count":49222,"following_count":822}},{"id":"96a7bc92-324a-4f73-a52d-5d86f1712685","tweet_url":"https://x.com/denk_tweets/status/2094799030757081434","tweet_text":"📝 Everyone claims that SaaS is dead. They're all wrong.\n\nYou may have heard of the SaaSpocalypse.\nThe thesis: with the rise of AI agents and coding tools, people and businesses are all going to build their own bespoke software, leaving these incumbent software businesses behind with an ever-diminishing customer base. That, and the rise of agentic workflows that will destroy the per-seat model that has grown so popular these past few decades.\nEvery thinkboi on X shares this vision, claiming that software as we know it is dead. And in the past year, the SaaSpocalypse has gone from fringe to consensus.\nThe markets have taken note. Shares in public SaaS companies have gotten crushed, despite many of those businesses performing arguably better than ever.\nLet’s look at HubSpot, for example. Over eight quarters HubSpot compounded revenue at ~20%, expanded margins nearly every quarter, went GAAP-profitable, and grew free cash flow more than 50% — hitting its 2027 margin target a year ahead of schedule.\nBut the stock is down ~55% from its 52-week high (bottomed out at 68%).\n \nThe market hates uncertainty. Sure, performance has been good (arguably excellent), but the acceleration of AI is so “unprecedented” that it’s impossible to bet on the status quo (i.e. HubSpot remaining relevant in an AI-first world).\nBetting on the SaaSpocalypse means betting on the thesis that all of these companies will soon be moot in a world where AI deploys custom software for everyone. That every SMB on earth is just one Claude Code session away from firing their entire stack, and that nobody will ever have to maintain any of it.\nI’m not betting against Anthropic (or AI), but I am here to tell you that the SaaSpocalypse is totally overblown and complete bullshit.\nMost of those claiming that SaaS is dead have never actually built anything at all. They aren’t in the arena using the latest tools. They’ve never built or run a business at scale. They probably think that Opus is some sort of peptide.\n \nLet me tell you why they’re all wrong. Actually, let me take it from the very top.\nHere are a few examples of SaaS companies: Shopify, Adobe, Zoom, Canva, Salesforce, beehiiv, etc. Software as a service (SaaS) is exactly what it sounds like — you pay someone to provide a service you don't have to build yourself.\nWhen you host your storefront on Shopify, you don’t need to build your own payment infrastructure, or host your own website, or think about sales tax across all 50 states, VAT in the EU, PCI compliance, chargeback disputes, or fraud.\nIn fact, they have a team of nearly 8,000 people who specialize in e-commerce, payments, checkout, fraud, logistics, and everything else to ensure your storefront is best positioned for success. They even routinely push updates to make the Shopify software better while you sleep.\nYou get best-in-class infrastructure for your online storefront, and no one on your team even thinks about it at all. From that perspective, the $39 per month you’re paying Shopify feels like a steal. That’s the beauty of SaaS.\nWe can use beehiiv as an example too. Technically, anyone has been able to build their own email infrastructure for decades. The same way anyone could build their own storefront for e-commerce. AI just makes it easier, and puts it within reach of people who aren't technical.\nWith the help of AI, sure, you could spend thousands of dollars and several months building your own email platform and have your own (much worse) version of beehiiv, custom-built just for you.\nBut when your team needs to integrate the newsletter with your podcast, or send dynamic content to your readers, or have access to enriched audience data… you’ll need to go back to the well and have someone on your team build these features themselves.\nWhen you want to monetize your newsletter, whether with ads or subscriptions, you’ll need to build your own sales pipeline, reporting, user authentication, payment rails, renewal logic, customer alerts, and more.\nWhen your emails start to land in spam, your team is now solely responsible for figuring out email deliverability all on their own.\nAnd remember — replacing SaaS and building the software yourself isn’t free. You still need to pay for the infrastructure, servers, maintenance, monitoring, and the other software and plugins required to make it all work. Not to mention the added human capital required to build and manage everything (indefinitely).\nSo after all of the time, money, and labor required to build a worse version of beehiiv, you’re now stuck having to maintain and update it all on your own forever. Or, hear me out, you could pay a small premium to have 130 specialists own all of that for you. That premium comes with regular software updates, access to the latest tools, uptime guarantees, and a team of experts in email deliverability and creator tools who can help you whenever you need it.\nThat’s the crux of the SaaSpocalypse debate. And there’s a small vocal minority of indie hackers who run solo teams claiming that everyone is going to build their own software just because it’s now technically possible.\nIf that were true, you would expect me (the CEO of a tech company, with 50+ engineers and a huge bias towards building vs buying) to have totally replaced all of our SaaS spend. But we still happily pay for Slack, Sentry, Linear, HubSpot, Loom, Zendesk, Amplitude, Ramp, Notion, Customer.io, etc. In fact, we pay considerably more for these today than we did just a year ago.\nAs my COO, Dan, says: we’ve been able to bake bread forever. Most people still buy it.\nMeanwhile, I’m not just yapping from the peanut gallery talking hypotheticals either. I’m in the arena. I’ve done the work myself.\nLast week, I announced that a few friends and I launched a social wellness club in Medellín. It’s called MADRE, and it’s home to the largest sauna in all of Latin America. (By the way, if you live there you should book a session. And if you don’t live in Medellín, you should follow us on Instagram).\nWe could have used Mindbody, WellnessLiving, Walla, or any of the other platforms to manage bookings and memberships. It would have cost less than $100 a month and been ready to use in a few hours. SaaS baby.\nBut out of curiosity, watching this whole vibe coding movement from the sidelines on X… I wanted to take advantage of the opportunity to get my hands dirty and build it myself.\nAs for my credentials: I'm a self-taught developer. I led product and engineering at Morning Brew, was a Product Lead at YouTube, and currently run product at beehiiv. I've built a lot of software, and I think I have a decent eye for design.\nHere’s the MADRE landing page to vouch for that design eye 😉\n \nKeep all of that in mind as I explain what happened next…\nI spent about 6 full weekends, I’d estimate maybe 80+ hours, building this website and platform from scratch. I probably spent another 20+ hours doing QA to ensure everything works perfectly.\nI had to build custom integrations with Twilio, Resend, Customer.io, Vercel, Slack, Sentry, Supabase, and Wompi (a local payment provider in Colombia). Yes, pay for and configure each of those individual accounts as well. I built an admin portal so our managers can run the business. I had to purchase datáfonos (payment terminals) to build and test the physical payment infrastructure.\nI also had to create user guides and tutorials to hand off to the team so they could learn and operate the software on their own.\nAnd that’s just the upfront investment. When the staff provided feedback, I had to build new features to support their requests. I also have Sentry errors piping into Slack to alert the team whenever there’s an issue with the platform.\nAs for who has to update and fix the software when it’s broken? That’s on me. But I have a full-time job running beehiiv and don’t have the bandwidth to do that outside a couple of hours here and there on the weekends (uh oh).\nAdd up Twilio, Vercel, Customer.io, Sentry, Resend, and Supabase and we're paying multiples more than what a SaaS company would charge to handle all of it for us. And that SaaS company would be improving the product, watching it for problems, and answering the phone when something breaks.\nFor the low, low price of $300 per month, and hours of my life I'll never get back… we have a fully custom-built solution that we can edit and tweak however we'd like :).\nRemember my credentials? I've led product and engineering at multiple startups and it still took me a hundred hours. Hell, I’m still technically on-call to support it.\nDo you think the average person selling cowboy boots in Jackson Hole is going to cancel their $39 per month Shopify subscription to spend 100+ hours building their own store, and then hire someone to manage it forever?\nOf course not. And the millions of small businesses paying for Salesforce aren’t going to build their own CRMs either.\n \nI rest my case.\n\nhttps://x.com/i/article/2094562662445367296","summary":"The post demonstrates that despite AI making custom software development technically feasible, SaaS remains economically rational because the hidden costs of building and maintaining bespoke solutions—including infrastructure, integrations, ongoing maintenance, and opportunity costs—consistently exceed subscription fees, especially for non-technical businesses. For builders and founders, this means the SaaSpocalypse narrative is overblown; the real opportunity lies in improving SaaS products to justify their value proposition rather than betting on mass defection to custom-built alternatives, while AI practitioners should focus on enhancing existing platforms rather than assuming they'll become obsolete.","author_name":"Tyler Denk 🐝","author_handle":"denk_tweets","timestamp":"2026-09-01T14:46:23+00:00","comment_count":14,"like_count":40,"retweet_count":6,"submitted_at":"2026-09-01T14:50:35.205908+00:00","created_at":"2026-09-01T14:50:35.205908+00:00","seo_title":"SaaS is not dead despite AI coding tools","bookmark_count":29,"image_url":"https://pbs.twimg.com/media/HRFnP6nbsAAwCRT.jpg","keywords":["saas","artificial intelligence","product management","software engineering","entrepreneurship","venture capital","business","startups","growth","technology"],"video_url":null,"video_thumbnail_url":null,"image_grid_urls":null,"profile_id":"1bc9aa79-75fd-4425-a591-57a82fb021b6","link_card_url":null,"link_card_title":null,"link_card_image_url":null,"link_card_domain":null,"is_pinned":false,"pinned_at":null,"content_type":"article","article_title":"Everyone claims that SaaS is dead. They're all wrong.","article_image_url":"https://pbs.twimg.com/media/HRFnP6nbsAAwCRT.jpg","slug":"everyone-claims-that-saas-is-dead-they-re-all-wrong-2094799030757081434","quality_score":8,"noindex":false,"profiles":{"id":"1bc9aa79-75fd-4425-a591-57a82fb021b6","bio":"cofounder/ceo @beehiiv. former product at youtube, morning brew. creator of @bigdeskenergy","name":"Tyler Denk","handle":"denk_tweets","website":"mail.bigdeskenergy.com/bio","location":"Los Angeles / Medellin","verified":true,"avatar_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/avatars/denk_tweets.jpg","banner_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/banners/denk_tweets.jpg","joined_date":"2012-01-01T08:00:00+00:00","tweet_count":11908,"listed_count":341,"follower_count":44964,"following_count":25009}},{"id":"6b9f5e52-91de-4be7-b593-a19ce7e56555","tweet_url":"https://x.com/poteto/status/2094457600259842065","tweet_text":"📝 The Complete Guide to pstack Pt. 1\n\nIn this series of posts, I'm going to show you how I use pstack, my personal set of skills for doing rigorous engineering work. It's allowed me to ship 2,000 PRs a month to production with high confidence. \n \nPersonally, I have never put much emphasis into how many lines of code or how many PRs I was landing. Before agents, no one cared, and rightfully so, as raw productivity did not always equate to quality or a visible outcome for users. It was simply a vanity metric.\nBut I've discovered through the course of building pstack that volume does matter, especially when you are able to maintain or even increase the level of quality of the product with agents. For example, I started working on Grok @Bot about 2 months ago, when it was still in its early days and the codebase was fresh but starting to grow. Despite the team growing and now landing hundreds of PRs a day into the Grok @Bot codebase, pstack has allowed me to keep the quality of the code high for everyone as I constantly monitor code, refactor, add new lints and checks, and also work on features. \n \nBeing Grok @Bot's gardener and maintainer is something I was only able to do through pstack. Our early momentum after building the prototype was very high and many people were joining the team. I had a critical moment of opportunity to refactor the whole codebase, while it was being built and extended and with no downtime, into something with strong foundations. A codebase with high quality that scales no matter how many engineers (and most importantly, non-engineers) contribute to it. All of this work requires me to improve the foundations of Grok Bot as it's being built, and you can only do that when the foundations can keep up with the number of contributions. Volume does matter now.\n \nThe proof is in Grok @Bot itself. Over the next few weeks, I'll tell you everything you need to know to be able to build and maintain a high quality app using pstack.\nPart 1 – Verification is all you need\nThe most critical skill to have in your toolbox is a high quality verification skill. This skill is so important to have and maintain that I think of it more like critical infrastructure rather than \"just\" a skill. A good one will amplify the output of your whole team, including non-engineers. Done well, you will 100-1000x your whole team's output. \nIf you're not familiar with the term, verification means that an agent can verify its own work. It can keep going until it succeeds at its task, because it can now close the loop without you being the bottleneck. If you're interested to know more of the story of how I created my first verification skill for Cursor, check out my previous post Loops You Can Trust.\nLet's build a verification skill together\nTo start, install pstack and then run /create-verification-skill. I also recommend adding Dr Eggbot, my bot that helps you create high quality bots, to your roster. Dr Eggbot comes with pstack which it will teach coding bots how to use, and non-coding bots that have similar rigor.\nYou can ask Dr Eggbot to create an engineer bot for you that you can then ask to run /create-verification-skill and setup a daily routine to run /maintain-verification-skill. \n \nWhile that runs, let's walk through what the skill does and how it makes a high quality verification skill for you.\nI distilled all of our verification skills that we use to build Grok @Bot and Cursor into this skill as a sort of meta-skill. It teaches your agent how to create a high quality one for your own app. \nNow this is where choice of tech stack is important. If you're building an app in Electron or for the web for example, you can take advantage of the rich debugging tools available for the JS ecosystem. For example, the Chrome DevTools Protocol (CDP) allows you to use the same tooling available in your browser's developer tools. Or if you're building an iOS app, making use of the simulator.\nYou ideally want the ability to interact with your app, debug it, take perf traces, and any other debugging and development tooling that you might typically use if you were developing the app by hand. If you don't have a rich runtime to make use of, you may need to ask your agent to create tools for you (eg using lldb, or a custom package that runs as a sidecar in dev environments), or just make use of what you have available.\nI personally feel that agentic verification is so important that I would unironically suggest building your own rich debugging tools, or even choosing a different tech stack, in order to have unfair advantages and extreme productivity in building software. As I mentioned earlier, giving agents the ability to verify their own work unlocks everyone in your organization to be able to contribute and validate that their changes actually work. The harder your tech stack is to debug and control, the more difficult it will be to use agents productively.\nMake it Reproducible\nIn pstack, we have a principle called \"Build the Lever\". What this means in the context of creating a skill, is that we prefer to give agents tools rather than just markdown. For verification skills, this means creating a small CLI that scripts interaction and debugging of your app in a small, agent friendly utility. This means that agents consume fewer tokens trying to do a task (run a CLI command instead of writing a throwaway script to click on something), and makes your verification skill more reproducible and testable. \nHere's a hypothetical example of a CLI your agent might make for an Electron app:\n \nNow, all agents can use this CLI to quickly navigate and debug your app. You'll also want to start thinking about the dev experience of building your app: \nseeding a dev database\nhow to handle auth, test users, API calls against a test/staging environment\ninstalling and bringing up your dev environment in a consistent way\nAll of this is stuff you've probably needed to think about anyway when you were writing code yourself. So think of this as your agents' main utility for doing dev work on your app. Keep it well maintained and tested!\nSome other example commands you might want to consider: \n \nOnce you have this basic setup, you should already start to see a big improvement in your agents. They should be able to navigate around and debug your app with ease.\nI recommend spending time here making this CLI good and error free before doing anything more advanced. You'll also want to think about (or ask your agent to) designing an agent friendly CLI. There are many resources online you can point your agent to, but the key properties I like are:\nthe API is easy to compose - think John Ousterhout's deep modules philosophy\nany command with potentially destructive side effects should have a --dry-run option\nmake use of subcommands to gradually disclose functionality rather than all at once\nerror messages should be very descriptive and tell the agent what it should do instead\nrich --help text\noutputs returned in machine readable form (eg JSON)\nGo faster with parallelism with Cloud Agents instead of worktrees\nWhen you've had some success running your verification skill to land a few PRs, you might start to wonder if you can parallelize more. For example, if an agent can now take your prompt and mostly drive it to a mergeable state, doesn't that free you up to run more agents? \nYour first instinct will be to add worktree support, meaning that your agents can use git to create a tracked copy of the repo where they can make changes in isolation to the main checkout. In theory, this lets you run multiple agents at once without their changes clobbering over each other.\nI would recommend against doing this. For one, it uses a lot of storage space and resources on your machine. You may be able to get away with running up to 10 agents in parallel with worktrees depending on the size of your repo and how powerful your machine is. But there's a far better way! \nCursor's cloud agents are agents that run on the cloud, on Cursor's infrastructure. These agents have access to a real computer, meaning that they can install dependencies, run your app, take videos and screenshots, and interact with your app like a real user can. If you've invested enough in the previous step to make your dev experience good, it shouldn't be a huge lift to be able to setup cloud agents. When you first setup your cloud environment, we send an agent to help you get it setup and running correctly. After the first build, we take a snapshot which means that subsequent cloud agent runs always start up quickly.\nI highly recommend taking the time to set cloud agents up, as it unlocks a massive increase in productivity in parallelism. In a later post I'll show you how I run hundreds of subagents in parallel in the cloud! But for now, setup your environment and get it to a state where you can start to feel confident about running all your agents in the cloud.\nKeep agents smart with Feature Maps\nAs your app grows more complex, agents need more guidance to be able to find features and interact with it. To do this, I've come up with something I call the Feature Map. As the name suggests, it's an easily searchable map of all the features available in your app, what it does, and how to get to it from a user's perspective. \nHere's an example Feature Map that I've prepared for a fictional app called Atlas. It's just a couple of markdown file that's mentioned in the verification's SKILL.md. \nYou can put this file anywhere, but in /create-verification-skill we automatically create a references/features directory alongside a README.md. The readme is the map itself: a high level overview of all the major features available, with links to specific details. An example feature looks something like this:\n \nDon't worry about writing these yourself! When you run /create-verification-skill, your agent will automatically go through your app and catalog everything and create these references for you. \nThe Feature Map, when combined with the CLI, is one of the main reasons why pstack's verification skills are so good. Agents now have context about every single feature and how to get to it, saving precious tokens in its context window and teaching it exactly what it's for and how to get there.\nYou can think of the Feature Map as a form of \"materialized memory\". If you've been using agents for a while you're probably familiar with the concept of memory - typically these might be stored as simple markdown files (eg an Obsidian vault), or even something more complex like a vector database. Personally, I think your codebase is the ultimate form of memory. Code is a projection of the decision making you and your team have made and represents the source of truth for what's happened and how things actually work. A Feature Map is just a more compact form of that, designed to save tokens. And because it's just markdown inside of a skill, everyone contributing to your codebase benefits from this shared memory. \nThis means that maintaining the verification skill is really important. I recommend running /maintain-verification-skill at least once a day to ensure that your agents always have the latest details on controlling your app. You may also find, as you use your verification skill more, that agents will automatically update them as they work on your app. /maintain-verification-skill catches whatever is missed. \nHow to use your verification skill\nFor reference, here's an example verification skill created for a fictional app: https://github.com/poteto/verification-skill-example. As a reminder, run /create-verification-skill to make one, which includes a basic CLI and Feature Map.\nHere's how I typically use it with pstack. \nFirst, of course, is to start your prompt with /poteto-mode. If you're using pstack through Cursor, you can also hit Opt + Enter instead of just Enter when you autocomplete /poteto-mode - this adds the skill as a Custom Mode, which pins the skill so your agent gets a reminder to use the skill on every new turn. \n \nIn Grok @Bot, install the plugin, then type /poteto-mode.\n \nExample: Building new features\nFor building new features, I typically use the verification skill alongside /poteto-mode to get the agent to verify its work. For example, I might prompt something like:\n/poteto-mode build <description of feature, any useful context>. use /control-app to verify your changes and show me a video and screenshots as proof\nIn Grok @Bot, I would prompt something like:\nspawn a cloud agent to use /poteto-mode to build <description of feature, any useful context>. use /control-app to verify your changes and show me a video and screenshots as proof\nThe minor difference here is that in Grok @Bot you tell your bot to spawn a cloud agent instead of doing the work itself. The main reason I prefer to do this is because it frees up your bot to do other things and keeps its context window clean. In that sense, I think of my bots more as coordinators who manage and supervise cloud agents. Cloud agents also mean that you can take advantage of the full array of models available in Cursor which have their own separate machine, so your bot's computer stays free for other things.\nExample: Perf work\nspawn a cloud agent to use /poteto-mode to improve the initial loading time of our app. first use /control-app to take a trace of the status quo, and identify opportunities for improvement. then do a targeted fix and use /control-app + a /swarm to confirm the win\n/swarm is one of the best skills to combine with your verification skill. It fans out any number of cloud agents to run your verification skill, so you can do things like confirm a perf win with a big enough sample size, or fuzz your app to ensure you didn't break or regress anything.\nInvest in your verification skill\nOnce you've created your verification skill, keep it sharp with /maintain-verification-skill. Keep improving the CLI and invest in the skill like you would critical infra. You may even want to put an oncall rotation on it - that's how important it is to unlocking 100-1000x productivity for your team.\nThis skill is the foundation for many other skills that we'll cover in the pstack guide, and composes beautifully with all of them.\npstack: https://x.ai/bot/plugin/9717366\nDr Eggbot: https://x.ai/bot/93gOz3op1UQdBdbekQFLK\nThanks for reading and stay tuned for Part 2!\n\nhttps://x.com/i/article/2094151284949688320","summary":"Understanding pstack's verification-first approach matters because it demonstrates how to architect AI agent workflows with self-closing feedback loops that scale team productivity 100-1000x, enabling non-engineers to contribute reliably while maintaining code quality across hundreds of parallel PRs. For builders using agents, this reveals that tech stack choice, debugging tool richness, and reproducible CLI design are now strategic infrastructure decisions that directly determine whether agents can autonomously validate their work, making verification skills the foundational lever for transforming volume-based shipping into sustainable, high-confidence delivery.","author_name":"lauren","author_handle":"poteto","timestamp":"2026-08-31T16:09:39+00:00","comment_count":113,"like_count":3087,"retweet_count":261,"submitted_at":"2026-08-31T17:10:06.125039+00:00","created_at":"2026-08-31T17:10:06.125039+00:00","seo_title":"Maintaining Code Quality While Scaling Agent Contributions","bookmark_count":5879,"image_url":"https://pbs.twimg.com/media/HQ_0MA_aMAA9zMt.jpg","keywords":["software engineering","code quality","agent verification","productivity","testing","debugging","continuous integration","developer tools","automation","machine learning agents"],"video_url":null,"video_thumbnail_url":null,"image_grid_urls":null,"profile_id":"4f28b977-4fdb-45de-85c8-609b2e63709f","link_card_url":null,"link_card_title":null,"link_card_image_url":null,"link_card_domain":null,"is_pinned":false,"pinned_at":null,"content_type":"article","article_title":"The Complete Guide to pstack Pt. 1","article_image_url":"https://pbs.twimg.com/media/HQ_0MA_aMAA9zMt.jpg","slug":"the-complete-guide-to-pstack-pt-1-2094457600259842065","quality_score":8,"noindex":false,"profiles":{"id":"4f28b977-4fdb-45de-85c8-609b2e63709f","bio":null,"name":"lauren","handle":"poteto","website":null,"location":null,"verified":false,"avatar_url":null,"banner_url":null,"joined_date":null,"tweet_count":4,"listed_count":0,"follower_count":0,"following_count":0}},{"id":"24df487a-c707-4ff0-9661-35ab3bcedb57","tweet_url":"https://x.com/a16z/status/2094437130810384556","tweet_text":"📝 Expanding the a16z Growth Fund and Platform\n\nBy @DavidGeorge83 and @RaghuRaghuram\nThere has never been an investing environment like what we’re living through right now. Normally, investors hope and pray to catch the beginning of one generational S-curve of technology adoption. Today we’re watching at least six mega-trends emerge:\nEnterprises are putting AI tools to work across every part of their organizations. We believe there is no upper bound on the demand to turn compute into business outcomes.\nConsumer AI is barely even a thing yet, aside from ChatGPT’s form factor beginning to replace the search engine. But we expect it will be transformational when the time is right.\nAmerican Dynamism has entered a new era. National security, industrial policy, and technological progress are converging to rebuild the physical systems that underpin the country, from defense and manufacturing to energy, infrastructure, and space.\nRobotics and autonomy are poised to be the biggest distributed infrastructure deployment in our lifetime, transforming how people and products move and interact in every setting, in our factories, on our roads, and everywhere in between.\nHealthcare is 18% of GDP and represents one of the biggest areas of opportunity, as our biology becomes programmable and we use AI to improve systems.\nTo power the above, the entire compute stack is being rebuilt for the AI-era.\nIn moments of opportunity like these, we keep coming back to the founder as the key to building generational businesses. Technology has become so powerful for company-builders, and yet so unevenly diffused in the economy, killer founders who understand it will always be the premier destination for investment capital. Our job as investors is to identify these founders and help them however they need.\nToday, a16z Growth has closed additional capital, bringing our fifth Growth fund to a total of $8.5B. With so much opportunity in the market right now, this became the obvious thing to do for both founders and our Limited Partners.\nHelping founders accelerate through the curve\nEvery founder who’s been through the scaling journey knows that there are inflection points where their early instincts aren’t enough anymore. Companies must become multi-product, multi-channel, multi-geography, often all at the same time. Databricks had to bank through the corner and evolve “Lakehouse” into a dominant platform. SpaceX evolved from a launch provider into a global communications, data infrastructure and AI company. Greatness isn’t easy.\nThe founder’s opportunity, at the growth stage of the business, is to keep growing really aggressively through each corner. In our 7+ years running a16z Growth, we’ve helped over 100 companies through these transitions.\nThe a16z Growth Platform: built by operators, for operators\nThe a16z Growth Platform includes access to a16z’s comprehensive offering of expertise, network, and infrastructure tailored to growth-stage companies. Our GTM team connects portfolio companies with new customers; Talent has built one of the world’s best executive and technical talent networks; Global efforts unlock international relationships for capital, partnerships, and customers; and New Media helps our founders and companies brand build.\nAnd now, we’re excited to offer even more ways to help our founders grow at scale, with increased offerings across sales, marketing, and pricing. Specifically, the team will be helping with:\nSales & Marketing Leadership: Defining what “great” looks like; helping build the right team and structure for the next stage of growth.\nAI-Native GTM: Helping companies compete with new playbooks for agentic operations, AI-native RevOps, demand generation, and consumption-based pricing.\nGTM Strategy & Positioning: Defining ideal customer profiles, category narrative, and core messaging; aligning sales and marketing on go-to-market strategy, and ensuring budgets back the right priorities.\nScaling Sales & Marketing Motions: Moving from founder-led sales to revenue engines, with the playbooks, campaigns, and metrics that drive predictable growth.\nPricing & Packaging: Navigating pricing model transitions and packaging decisions as companies move from single-product to platform, including AI margin governance.\nRevenue Operations: Designing comp plans, managing usage-based pricing transitions, and building the AI-native operations that keep pace with growth.\nThe a16z Growth Platform is made up of elite operators who understand the challenges of growth-stage companies and how to execute in their respective domains. They bring actual game footage to every interaction with founders and CXOs. The team upleveling our sales, marketing, and pricing efforts is no different, having spent time in-house or closely advising companies such as Lovable, Atlassian, Samsara, 1Password, Miro, PagerDuty, Segment, Workday, and others during their hypergrowth and pre-IPO years.\nIt’s time to grow\nTop operators today are in the biggest job of their lives, in the fastest-moving, most contested market we’ve ever seen. We know it takes both capital and operational know-how to win, and we are building to meet these demands.\n\nhttps://x.com/i/article/1762991355478061056","summary":"a16z's $8.5B Growth Fund expansion directly addresses the operational bottleneck that growth-stage founders face: scaling across multiple products, geographies, and channels requires specialized expertise beyond capital, which the fund now provides through dedicated teams in sales, marketing, pricing, and AI-native revenue operations. For builders and AI practitioners, this signals both the market validation of six converging mega-trends (enterprise AI, consumer AI, American Dynamism, robotics, healthcare, and compute infrastructure) and a new playbook for hypergrowth that emphasizes operational excellence alongside funding, making founder-friendly capital increasingly competitive based on platform value rather than check size alone.","author_name":"a16z","author_handle":"a16z","timestamp":"2026-08-31T14:48:19+00:00","comment_count":14,"like_count":185,"retweet_count":16,"submitted_at":"2026-08-31T15:25:05.846321+00:00","created_at":"2026-08-31T15:25:05.846321+00:00","seo_title":"Andreessen Horowitz expands Growth Fund to $8.5 billion","bookmark_count":65,"image_url":"https://pbs.twimg.com/media/HRDuXhIWAAEbRwq.jpg","keywords":["venture capital","growth marketing","saas","artificial intelligence","ai tools","robotics","entrepreneurship","fundraising","go-to-market strategy","revenue operations"],"video_url":null,"video_thumbnail_url":null,"image_grid_urls":null,"profile_id":"3343945f-a1d4-42e1-bb13-d1af36a62ab5","link_card_url":null,"link_card_title":null,"link_card_image_url":null,"link_card_domain":null,"is_pinned":false,"pinned_at":null,"content_type":"article","article_title":"Expanding the a16z Growth Fund and Platform","article_image_url":"https://pbs.twimg.com/media/HRDuXhIWAAEbRwq.jpg","slug":"expanding-the-a16z-growth-fund-and-platform-2094437130810384556","quality_score":6,"noindex":false,"profiles":{"id":"3343945f-a1d4-42e1-bb13-d1af36a62ab5","bio":"we invest in software eating the world \nhttp://a16z.com/portfolio\nhttp://a16z.com/podcasts\n\nWatch \"The Ben & Marc Show\": https://youtube.com/@a16z","name":"a16z","handle":"a16z","website":"a16z.com","location":"The Cloud","verified":true,"avatar_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/avatars/a16z.png","banner_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/banners/a16z.jpg","joined_date":"2009-08-01T07:00:00+00:00","tweet_count":25687,"listed_count":13791,"follower_count":1054053,"following_count":64}},{"id":"70b68fe3-828d-4620-a278-89b12920ec8e","tweet_url":"https://x.com/gregisenberg/status/2093783102438928873","tweet_text":"📝 I lost $5M in startups (i'll tell you the real story)\n\nIn 2013, at 23, I sold my company.\nIt got rolled into a bigger venture-backed company, and my slice was worth about $5 million on paper. Houses in my town cost $290k back then, so in my head I was set for life.\nYears later, that bigger company sold too. I got an email saying it had sold (yay!) but that common stock holders were getting $0 (boo!).\nSee, my $5 million was common stock, and that company was doing $35M in revenue at 70% margins, backed by huge names. On paper I looked rich. But once the preferred investors got paid back first, there was nothing left for the rest of us. Gone. Zero.\n \nBest lesson I ever got, and I'm grateful I learned it at 23 instead of 43.\nHere's what it taught me. Startup equity is just gravy (and you can't count on gravy). Whether that paper number ever turns into money depends on a hundred things you are often out of control: the market, the timing, the next round, who got paid before you in the stack. You can do everything right and still open an email that says $0. \nEven founders who have built billion dollar companies before, sometimes can't do it again, again, again. \nSo I build differently now. Profitable. Steady. No venture capital. Twice a year I send the team profit shares, and the reaction is always the same. \"Wait, this is real?\" It's real because it's cash in a bank account, not a number on a cap table waiting for the stars to align. \nExamples of some of my businesses. There's LCA, our design firm building agentic interfaces for companies like Slack, Intuit, and Character AI.  This business is massive. Our team has 5+ years building intelligent products for the biggest brands in the world. DM me if you want to work with us. \nAnd ideabrowser, the #1 tool for people looking for startup ideas, trends in the agentic era. Literally free to sign up for ideas so might as well. We'll be announcing more monetization soon.\nDifferent products, same engine: profitable, run lean with AI on the inside, and grown through creator-led distribution instead of a raise or a sales team. That's the whole holdco model.\nAnd to be clear, I'm not saying I'd never raise, or that VC is bad. I've seeded multiple unicorns and done really well from venture. It's a phenomenal machine for the right business at the right time for the right founder. I'm just saying it's one game, not the only one, and for most people it's not even the one that fits.\nAnd here's why this matters now more than ever in this agentic era....\nAgents just deleted the 2 reasons you used to need a war chest: people and time. One person points an agent at the support inbox and it clears the tickets. Another runs the whole paid acquisition account overnight. The research, the ops, the tools you'd have hired a team for, all of it runs on a few hundred dollars of tokens a month instead of a payroll. So the math starts looking really good. Basically, margins are about to go up and burn rate lower. Oh, and there is more opportunity, niches etc than ever \nPeople always ask me what they'd actually build. So here's the version I'd chase if I were starting today. \nPick a boring industry still running on Excel and email, specialty insurance, equipment leasing, commercial real estate. Build small agents that each handle one piece of the workflow and chain them together. Then post 60-second videos solving one real problem for that industry every day, end each with a free tool as a lead magnet, and when one clears a 2% click rate, put ad spend behind it. Organic gets you to a few hundred K, paid takes you to a few million, a webinar funnel on top gets you to $5M+. \nThis is high level how I'd think of it. Find super niches, find pain points, build agents, automate them, build a media machine, paid ads to supercharge, events/community/network effects for lock in etc. If you can build a business generating $2-3 million in profit, you can raise or can exit that business for $10M+. And you own all of it.\nOne honest caveat: this is hard too. Different hard, but hard. Nobody spins up a few agents and gets rich overnight, and most of these will still fail. I just think if you're driven and you keep swinging, your odds beat the VC path.\nAnd obviously, money isn't the only reason to build. But if you grew up without it, or you're trying to lock in a nest egg while actually enjoying your life, this matters.\nThere are two games now. One is raising a billion and swinging to be the next Sam Altman, where the odds are a rounding error. The other is running something with agents, like Pieter Levels, $3M+ a year, zero staff, from a laptop. \nNo right or wrong game. Just the right one for you.\nLosing that $5 million felt like the worst thing that could happen at 23. It might've been the best. It showed me which game I actually wanted to play. \nWhatever you pick, I'm rooting for you.\nI figured I'd share this story because why not.\nFeel free to share your versions. I hope it helps founders even just a little.\nLeast we can do.\nPay it forward.\n\nhttps://x.com/i/article/2093776326561873920","summary":"This post matters because it reveals that startup equity is inherently risky regardless of apparent success—common stockholders can receive nothing despite massive revenue and valuations if preferred investors are paid first—making it crucial for builders to understand cap table hierarchy and consider alternative paths like profitable bootstrapping. AI practitioners and founders should recognize that autonomous agents now make lean, profitable businesses viable by eliminating traditional burn rate requirements (replacing team payroll with token costs), fundamentally shifting the risk-reward calculation toward sustainable, creator-led businesses over venture-dependent scaling.","author_name":"GREG ISENBERG","author_handle":"gregisenberg","timestamp":"2026-08-29T19:29:26+00:00","comment_count":18,"like_count":136,"retweet_count":3,"submitted_at":"2026-08-29T19:45:18.194778+00:00","created_at":"2026-08-29T19:45:18.194778+00:00","seo_title":"How startup equity can disappear even with massive revenue","bookmark_count":195,"image_url":"https://pbs.twimg.com/media/HQ6a4ilWUAAyCmA.jpg","keywords":["venture capital","startup equity","profitable business","saas","entrepreneurship","ai agents","growth marketing","product design","side hustles","machine learning"],"video_url":null,"video_thumbnail_url":null,"image_grid_urls":null,"profile_id":"c6048ebb-f903-4de9-b02d-15ed9266e48b","link_card_url":null,"link_card_title":null,"link_card_image_url":null,"link_card_domain":null,"is_pinned":false,"pinned_at":null,"content_type":"article","article_title":"I lost $5M in startups (i'll tell you the real story)","article_image_url":"https://pbs.twimg.com/media/HQ6a4ilWUAAyCmA.jpg","slug":"i-lost-5m-in-startups-i-ll-tell-you-the-real-story-2093783102438928873","quality_score":6,"noindex":false,"profiles":{"id":"c6048ebb-f903-4de9-b02d-15ed9266e48b","bio":"I run a portfolio of internet companies and host @startupideaspod. CEO: @latecheckoutplz we build companies like @ideabrowser, @meetLCA, @boringmarketer etc","name":"GREG ISENBERG","handle":"gregisenberg","website":"gregisenberg.com","location":"more →","verified":true,"avatar_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/avatars/gregisenberg.jpg","banner_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/banners/gregisenberg.jpg","joined_date":"2008-05-01T04:00:00+00:00","tweet_count":63799,"listed_count":11222,"follower_count":692554,"following_count":996}},{"id":"c61af702-2e15-4c90-a666-3707351a0c21","tweet_url":"https://x.com/ChromiumDev/status/2093424350036660456","tweet_text":"📝 Chrome 153 beta\n\nUnless otherwise noted, the following changes apply to the newest Chrome beta channel release for Android, ChromeOS, Linux, macOS, and Windows. Learn more about these features by using the provided\n\nhttps://x.com/i/article/2093423740839153664","summary":"Chrome 153 beta introduces new platform capabilities and APIs that builders need to evaluate for web application development across multiple operating systems and devices. Understanding these features early in the beta cycle allows founders and AI practitioners to design applications that leverage emerging web standards and plan technical roadmaps before general availability.","author_name":"Chrome for Developers","author_handle":"chromiumdev","timestamp":"2026-08-28T19:43:53+00:00","comment_count":3,"like_count":53,"retweet_count":8,"submitted_at":"2026-08-28T19:45:03.966208+00:00","created_at":"2026-08-28T19:45:03.966208+00:00","seo_title":"Chrome 153 Beta Introduces New Features Across Platforms","bookmark_count":6,"image_url":"https://pbs.twimg.com/media/HQ1WNUSWAAAU0Uy.jpg","keywords":["web development","browser technology","software engineering","programming","chromeos","cross-platform development","developer tools","tech innovation"],"video_url":null,"video_thumbnail_url":null,"image_grid_urls":null,"profile_id":"eda3383a-d369-4f3c-b476-86e774ef0676","link_card_url":null,"link_card_title":null,"link_card_image_url":null,"link_card_domain":null,"is_pinned":false,"pinned_at":null,"content_type":"article","article_title":"Chrome 153 beta","article_image_url":"https://pbs.twimg.com/media/HQ1WNUSWAAAU0Uy.jpg","slug":"chrome-153-beta-2093424350036660456","quality_score":6,"noindex":false,"profiles":{"id":"eda3383a-d369-4f3c-b476-86e774ef0676","bio":null,"name":"Chrome for Developers","handle":"ChromiumDev","website":null,"location":null,"verified":true,"avatar_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/avatars/chromiumdev.jpg","banner_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/banners/chromiumdev.jpg","joined_date":null,"tweet_count":12076,"listed_count":5580,"follower_count":422295,"following_count":128}},{"id":"d321b330-7a7b-4798-a619-31842e6eeda5","tweet_url":"https://x.com/AndrewYNg/status/2093388974194872781","tweet_text":"📝 AI Engineering Skills Map: Software engineering fundamentals \n\nHow have software engineering fundamentals changed with agentic coding? Even when you use a coding agent to write all your code, understanding software fundamentals is important for steering your agent to make the tradeoffs you want — or to even know what tradeoffs exist to be made. Additionally, when you’re building an AI application, the AI core is often expressed through a broader software application, which you will want to help build or shape.  \nA novice who vibe codes without understanding software fundamentals can create simple applications, but this often leads to the coding agent making bad tradeoffs in latency, availability, consistency, reliability, maintainability, simplicity, and/or cost. In such cases, the developer didn’t know such tradeoffs even existed and therefore did not steer the agent to make the right decisions for their application context.\nThis article describes what our study of AI Engineering Skills shows are the most important things to know in software engineering. It requires being skilled at:\nBuilding full-stack applications\nManaging data\nDesigning system architectures\nMaking systems secure and reliable\nScaling and operating in production\nBuilding full-stack applications. Agentic coding enables many developers who previously played more specialized roles (like front-end developer or mobile developer) to play a broader, full-stack role. A coding agent can help with parts of the development process that you might be less familiar with. However, understanding how the full stack actually works is important. Skilled developers understand the key components and concepts of front-end and back-end systems, including UI components, caching, page rendering, API choice and design, authentication, state and session management, asynchronous processing, data persistence, testing, security, and accessibility.\nManaging data. Data deserves special attention because it is a foundation that software is built on top of, that is relatively hard to change (even if agents help with migrations). When you know how to manage data, you can think through access patterns and use them to decide what to store and for how long. You can identify the right data models and select the appropriate storage types (such as relational tables, documents, key-value, or graphs) and infrastructure, which in turn affects speed, scalability, availability, reliability, and cost. You understand transactions, concurrency, and how to ensure your data is clean, consistent, and fresh. When needed, you can ensure proper privacy, governance, and compliance. You know how to manage the data lifecycle.\nAs an application evolves, you also know how to evolve the data architecture with it. Deciding how to manage data requires significant human-provided context. Your AI systems will get their own input context from your data source, so if data architecture is chosen poorly, the AI doesn’t know what it doesn’t know. This is why it takes skilled intervention from someone with the relevant context and skilled at AI engineering — you! — to set it right. How to build data infrastructure for agents — rather than only traditional software or humans — is also a rapidly evolving area, and you should continue to adjust your best practices as the field evolves.\nDesigning system architectures. When you understand the major components of the full stack of software and data, you are then better positioned to decide how to put the pieces together. Good system design requires understanding what the software is intended to do (how many users? how important is latency? how important is cost? etc.) so you can make choices about the application platform, the boundary between the frontend and backend, system decomposition, application state placement, and architectural granularity (monolith vs. microservices). You will also choose the stack (programming languages, runtimes, component/frontend/backend frameworks, data technologies) — sometimes by running experiments to evaluate options before settling on one.\nFurther, the right architecture is a moving target, depending on the phase of the project. The simple architecture you choose to build a quick prototype may not be the right architecture to build the first production system, and that too may change as the application scales. Making these decisions requires deep technical knowledge of both software components and the application context so you can design — and evolve — the architecture to make better tradeoffs.\nMaking systems secure and reliable. To build reliable systems, you should know how to develop testing strategies to verify the correctness of your system: What mix of unit tests and integration tests, what frameworks to use, and what level of coverage. You also know how to design around possible failures — how to handle failures (like an API hitting a rate limit), build in graceful degradation, and minimize the blast radius of failures. Additionally, rather than first writing software and then later figuring out how to secure it, the “shift left” movement is moving security work earlier in the lifecycle (to the left on a traditional project timeline). Just as all developers are moving toward becoming full stack developers, many developers are now also partly security engineers. You can now use AI tools to scan your code for vulnerabilities, check dependencies for supply chain injections, and examine your cloud configuration for attack surfaces. But doing this well still requires some knowledge of security.\nScaling and operating in production. To serve real users, you will have to know how to deploy your software to production. You will benefit from knowing how to execute the software development lifecycle (SDLC) which, in addition to building and testing, includes configuring the deployment environment, deciding on release strategy, applying deployment automation (CI/CD), and understanding infrastructure as a service (IaaS).\nOperating in production requires putting in place observability tools, setting alerts, and managing incidents. Lastly, to scale your application, you should understand the real load and know how to scale servers, load-balance, and adapt your data infrastructure (via sharding, indexing, replication) or make architecture changes to allow your system to adapt to scale. Finally, understanding coding best practices like version control, code reviews, dependency maintenance, and how to manage technical debt helps you keep evolving your system over time.\nCoding agents have changed how we build software, including software that does not contain any AI components. Some parts of coding knowledge — like memorizing coding syntax — are becoming obsolete. But developers who deeply understand how software works vastly outperform those who vibe code without understanding.\nUnderstanding software fundamentals (in addition to AI) also helps you figure out what software can and cannot do. This makes them important context for how you use coding agents and shape the build. I will discuss these in future posts.\n\nhttps://x.com/i/article/2093384274372419585","summary":"Understanding software fundamentals remains critical for AI practitioners using coding agents because agents cannot independently recognize important engineering tradeoffs in latency, availability, consistency, reliability, maintainability, cost, and security—only informed developers can steer them toward context-appropriate decisions. Builders and founders need deep knowledge across full-stack development, data architecture, system design, security, and production operations to ensure that agentic coding produces robust applications rather than systems with hidden technical debt and poor architectural choices.","author_name":"Andrew Ng","author_handle":"andrewyng","timestamp":"2026-08-28T17:23:19+00:00","comment_count":112,"like_count":3043,"retweet_count":441,"submitted_at":"2026-08-28T18:15:03.568914+00:00","created_at":"2026-08-28T18:15:03.568914+00:00","seo_title":"Software Fundamentals Remain Essential With Agentic Coding","bookmark_count":4481,"image_url":"https://pbs.twimg.com/media/HQ0yTnbaoAAuJH3.jpg","keywords":["software engineering","ai engineering","system architecture","full-stack development","machine learning","data management","software fundamentals","cybersecurity","production deployment","scalability"],"video_url":null,"video_thumbnail_url":null,"image_grid_urls":null,"profile_id":"3ecf579e-e33f-4b35-8ac3-36de3de7ad8e","link_card_url":null,"link_card_title":null,"link_card_image_url":null,"link_card_domain":null,"is_pinned":false,"pinned_at":null,"content_type":"article","article_title":"AI Engineering Skills Map: Software engineering fundamentals ","article_image_url":"https://pbs.twimg.com/media/HQ0yTnbaoAAuJH3.jpg","slug":"ai-engineering-skills-map-software-engineering-fundamentals-2093388974194872781","quality_score":7,"noindex":false,"profiles":{"id":"3ecf579e-e33f-4b35-8ac3-36de3de7ad8e","bio":"Co-Founder of Coursera; Stanford CS adjunct faculty. Former head of Baidu AI Group/Google Brain. #ai #machinelearning, #deeplearning #MOOCs","name":"Andrew Ng","handle":"AndrewYNg","website":"andrewng.org","location":"Palo Alto, CA","verified":true,"avatar_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/avatars/andrewyng.jpg","banner_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/banners/andrewyng.jpg","joined_date":"2010-11-01T07:00:00+00:00","tweet_count":2020,"listed_count":20240,"follower_count":1803269,"following_count":1104}},{"id":"7f3d8d0c-5fe8-4244-afc8-231e399bd53e","tweet_url":"https://x.com/a16z/status/2093324635890933761","tweet_text":"📝 The Machine Age Fund\n\nWe’ve raised $1.1B for a16z’s newest fund: the Machine Age Fund. It’s time to open the throttle and accelerate the physical buildout of AI: the strongest tool ever developed for solving problems and\n\nhttps://x.com/i/article/2093207144376012800","summary":"This $1.1B commitment from Andreessen Horowitz signals major institutional capital flowing toward AI infrastructure and physical deployment, enabling founders to pursue ambitious projects in robotics, manufacturing, and hardware that require significant upfront investment. For builders and AI practitioners, this validates the shift from software-only AI applications toward embodied AI systems and physical automation, creating new opportunities to develop tools that solve real-world problems at scale.","author_name":"a16z","author_handle":"a16z","timestamp":"2026-08-28T13:07:39+00:00","comment_count":50,"like_count":832,"retweet_count":103,"submitted_at":"2026-08-28T14:15:32.11901+00:00","created_at":"2026-08-28T14:15:32.11901+00:00","seo_title":"For builders and AI practitioners","bookmark_count":634,"image_url":"https://pbs.twimg.com/media/HQyQ2ksaAAE2AZM.jpg","keywords":["artificial intelligence","venture capital","machine learning","ai infrastructure","deep learning","innovation","entrepreneurship","saas","software engineering","technology"],"video_url":null,"video_thumbnail_url":null,"image_grid_urls":null,"profile_id":"3343945f-a1d4-42e1-bb13-d1af36a62ab5","link_card_url":null,"link_card_title":null,"link_card_image_url":null,"link_card_domain":null,"is_pinned":false,"pinned_at":null,"content_type":"article","article_title":"The Machine Age Fund","article_image_url":"https://pbs.twimg.com/media/HQyQ2ksaAAE2AZM.jpg","slug":"the-machine-age-fund-2093324635890933761","quality_score":8,"noindex":false,"profiles":{"id":"3343945f-a1d4-42e1-bb13-d1af36a62ab5","bio":"we invest in software eating the world \nhttp://a16z.com/portfolio\nhttp://a16z.com/podcasts\n\nWatch \"The Ben & Marc Show\": https://youtube.com/@a16z","name":"a16z","handle":"a16z","website":"a16z.com","location":"The Cloud","verified":true,"avatar_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/avatars/a16z.png","banner_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/banners/a16z.jpg","joined_date":"2009-08-01T07:00:00+00:00","tweet_count":25687,"listed_count":13791,"follower_count":1054053,"following_count":64}},{"id":"9e6552c3-e5e6-4f56-a726-f7f454829893","tweet_url":"https://x.com/gdb/status/2093021551855812842","tweet_text":"📝 A call for collective action on cyber defense\n\nWe have a limited window to strengthen cyber defenses.\nIn the coming months, AI-enabled cyber attacks will become far more widespread and sophisticated as models around the world become increasingly capable. The companies and public services our communities depend on — from hospitals to water treatment plants to the infrastructure that powers the internet — are at risk.\nToday’s AI advances are already giving defenders new ways to fix weaknesses that have accumulated for years. If we act decisively, we can use the defenders’ window to make our digital world much more secure.\nWe propose the following principles for a collective response:\nRecognize that status quo security won’t be enough. Longstanding bugs, excessive permissions, misconfigurations, insecure and unpatched software, weak authentication, and technical debt in legacy systems have left systems exposed. Security teams, particularly for critical infrastructure, have been historically under-resourced and need a surge in tools and resources.\nEmpower more defenders with cyber-capable AI. AI brings specialist skills to more defenders and makes core security tasks faster, cheaper and better. Sharing tools, practical knowledge, and verified fixes lets one organization’s work help protect many others.\nMobilize a collective response. Cyber capabilities are advancing worldwide, and that can be a net positive: no single company should control the future. It also means a global response is necessary, requiring new partnerships to raise security standards and find new solutions to emerging cyber threats.\nEach of us can reduce risk now. All organizations, cybersecurity companies, technology partners, governments, and AI frontier companies have an important role: accelerate defenders’ priorities with tools, funding, and hands-on support, especially for critical infrastructure organizations with limited budgets.\nHere’s what we think needs to happen next:\n01 Every organization\nMake cyber defense an immediate leadership priority. Raise your security standards and meet them with the urgency and coordination of an incident that takes precedence over everything except critical business operations. Fix the highest-risk weaknesses, verify results without disrupting essential services, and raise the security bar for what you buy, build, and deploy, including AI-generated code. Upgrade or replace systems to build in least privilege, strong access controls, and defense in depth. Use capable, lower-cost models for broad coverage, and apply frontier capabilities to the hardest problems. Where a system cannot be patched without disrupting essential services, apply and verify compensating controls.\n02 Cybersecurity companies and technology partners\nHelp lead the response to defend against sustained AI-enabled attacks, including testing defenses continuously against frontier cyber capabilities, strengthening existing tools with AI, and working with technology partners to close gaps now. Make AI-powered defense accessible and deployable for critical-infrastructure operators, with hands-on help to deploy tools and verify fixes, and collaborate with critical infrastructure supply chain manufacturers and system integrators to patch and issue interim guidance. Share threat intelligence and tested playbooks, and measure progress by how many organizations are protected, how quickly attacks are contained, and whether fixes work.\n03 Governments\nCoordinate cyber defense at local, national, and international levels. Strengthen existing government and industry channels to share actionable threat intelligence, prioritize the most serious risks, and coordinate incident response and recovery around the world. Fund cyber defense, starting with essential services that lack the staff or budget to act. Expedite the expansion of trusted access programs, especially for critical infrastructure supply chains, and broaden access to defensive capabilities for other defenders. Give hospitals, water utilities, and local governments access to capable defensive AI, authorized testing, and hands-on support through trusted security providers and partners. Impose costs on attackers.\n04 Frontier AI companies\nProvide responsible model access, significant funding, training, and hands-on support, especially for under-resourced critical-infrastructure defenders. Build observability and security tools, ensure agentic identities are traceable and accountable, and share best practices in continuous monitoring. Invest in authorized testing, private disclosure, and verified fixes, and share tools, playbooks, and credible threat assessments with governments, security partners, and open-source maintainers to strengthen preparedness, response, and recovery.\n \nWe can make the digital infrastructure we all depend on more secure.\nWe call on leaders across industry and government to bring the full weight of their technology, resources, and expertise to this effort. Put cyber-capable AI in the hands of defenders, starting with the teams protecting essential services. Fix the most dangerous weaknesses, verify the fixes, and share what works so others can build on it. Together, we can turn today’s AI advances into lasting improvements in security that benefit everyone. Let’s put them to work.\nhttps://openai.com/collective-cyberdefense\n\nhttps://x.com/i/article/2093011711712456704","summary":"AI-enabled cyber attacks will become significantly more sophisticated in the coming months, creating an urgent need for builders and founders to prioritize cyber defense as a leadership issue while fixing accumulated technical debt, weak authentication, and legacy system vulnerabilities. AI practitioners and frontier companies have a critical responsibility to democratize defensive AI capabilities—making secure-by-default tools, threat intelligence, and verified fixes accessible to under-resourced critical infrastructure operators like hospitals and water treatment plants that cannot otherwise afford comprehensive security upgrades.","author_name":"Greg Brockman","author_handle":"gdb","timestamp":"2026-08-27T17:03:19+00:00","comment_count":232,"like_count":2648,"retweet_count":421,"submitted_at":"2026-08-27T18:05:20.142502+00:00","created_at":"2026-08-27T18:05:20.142502+00:00","seo_title":"AI practitioners and frontier companies have a critical","bookmark_count":1294,"image_url":"https://pbs.twimg.com/media/HQvj0tOaAAAWrhf.jpg","keywords":["cybersecurity","artificial intelligence","cyber defense","critical infrastructure","ai-enabled attacks","threat intelligence","security leadership","machine learning","cyber threats","technology policy"],"video_url":null,"video_thumbnail_url":null,"image_grid_urls":null,"profile_id":"8420794a-2091-4f94-93a2-18e3ad7fa7f8","link_card_url":null,"link_card_title":null,"link_card_image_url":null,"link_card_domain":null,"is_pinned":false,"pinned_at":null,"content_type":"article","article_title":"A call for collective action on cyber defense","article_image_url":"https://pbs.twimg.com/media/HQvj0tOaAAAWrhf.jpg","slug":"a-call-for-collective-action-on-cyber-defense-2093021551855812842","quality_score":6,"noindex":false,"profiles":{"id":"8420794a-2091-4f94-93a2-18e3ad7fa7f8","bio":"President & Co-Founder @OpenAI","name":"Greg Brockman","handle":"gdb","website":"gregbrockman.com","location":null,"verified":true,"avatar_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/avatars/gdb.jpg","banner_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/banners/gdb.jpg","joined_date":"2010-07-01T07:00:00+00:00","tweet_count":6396,"listed_count":11315,"follower_count":1033450,"following_count":8}},{"id":"8c8a12ec-46bd-4959-910e-1fec6edc2d23","tweet_url":"https://x.com/ericzakariasson/status/2092982710465970425","tweet_text":"📝 How I run multiple teams of Grok Bots\n\nMost of the time I run my Grok Bots as separate chats. Coder in one, Writer in another, Researcher in a third. That's all fine until you need juggle couple of different projects at the same time. It becomes hard filter out signal from noise, and keeping them in sync.\nAs a human and a team of humans, we tend to coordinate work in projects. Each project gets a spec, plan, perhaps a Slack channel to communicate in. Then for each project, you'd have different tasks that someone takes on.\nSo, I replicated that. Each project gets a Grok Bot channel and an entry in a Notion database. Note that this is an experimental pattern I'm still trying out\n \nThe setup\nI created two databases in Notion, Projects and Tasks. Then, I created a projects (plural) Manager bot. This will take care of the higher level coordinate of projects and tasks. It has a /Project Ops skill that takes care of creating a project, opening a channel, and staffing it with the right bots. It's responsible for the meta work.\nOne project equals one channel in the sidebar, and the channel has the team of bots (and me) in it!\n \nThe roster lives in that channel.\n \nStaffing rules I added:\nReuse existing bots first. This can be a Coder, Researcher, Writer or anything else that I've already defined.\nPropose at most five bots besides the PM. Each channel is limited at six bots (just an arbitrary number I came up with)\nCreate a new bot only when nothing on the bench fits, and only after I say yes. This could be a specialist in a domain that gets reused in the future, or en ephemeral one only for this project.\nHere's what the Projects database looks like\n \nHow I use it\nOnce the channel exists, I talk in it with the roster and we scope out the tasks for the project.\nThe PM sits in the channel too, but mostly watches progress and makes sure the databases are up to date.\nWhen a bot gets stuck, or needs more input, it'll mark its task as Blocked and then ping me in the channel. Often the bots can accomplish a lot on their own, and I can just watch the cards getting move around. Quite satisfying to be honest.\n \nThe interesting part is that the more I build on this, the more it resembles a system initially built for humans. A board, a manager, specialists claiming tasks, a blocked column, a channel.\nThat is the experiment so far. If you run a team of bots, I want to see your setup! How do you run yours?\n\nhttps://x.com/i/article/2092982704329605120","summary":"This matters because it demonstrates a practical organizational pattern for scaling AI agent workflows beyond single-task chatbots—using project channels, task databases, and a manager bot to coordinate multiple specialized agents with human oversight, which directly translates to how builders and founders can structure multi-agent systems for complex product work. The staffing rules and reuse-first approach provide actionable constraints that AI practitioners can apply to avoid bot proliferation while maintaining efficiency, showing that human team management principles effectively structure AI agent coordination when adapted properly.","author_name":"eric zakariasson","author_handle":"ericzakariasson","timestamp":"2026-08-27T14:28:58+00:00","comment_count":54,"like_count":873,"retweet_count":55,"submitted_at":"2026-08-27T15:10:11.62059+00:00","created_at":"2026-08-27T15:10:11.62059+00:00","seo_title":"Running Multiple Grok Bots With Project Management Structure","bookmark_count":2034,"image_url":"https://pbs.twimg.com/media/HQvEo3dbQAAjAG3.png","keywords":["ai workflow management","project management","team coordination","automation","bot orchestration","productivity systems","remote work tools"],"video_url":null,"video_thumbnail_url":null,"image_grid_urls":null,"profile_id":"2cf6f0f2-2efd-48b9-ad6a-0ea33110c537","link_card_url":null,"link_card_title":null,"link_card_image_url":null,"link_card_domain":null,"is_pinned":false,"pinned_at":null,"content_type":"article","article_title":"How I run multiple teams of Grok Bots","article_image_url":"https://pbs.twimg.com/media/HQvEo3dbQAAjAG3.png","slug":"how-i-run-multiple-teams-of-grok-bots-2092982710465970425","quality_score":6,"noindex":false,"profiles":{"id":"2cf6f0f2-2efd-48b9-ad6a-0ea33110c537","bio":"@cursor_ai & tinkering. http://colf.dev","name":"eric zakariasson","handle":"ericzakariasson","website":"anyblockers.com","location":"San Francisco, CA","verified":true,"avatar_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/avatars/ericzakariasson.jpg","banner_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/banners/ericzakariasson.jpg","joined_date":"2012-02-01T08:00:00+00:00","tweet_count":6783,"listed_count":929,"follower_count":82481,"following_count":601}},{"id":"e9c56959-8403-4551-a618-9eecaf2567a1","tweet_url":"https://x.com/ChromiumDev/status/2092690739130196410","tweet_text":"📝 New in Chrome 152\n\nChrome 152 is rolling out now, and this post shares some of the key features from the release. Read the full Chrome 152 release notes.\nHighlights from this release:\nCSSPseudoElement support for additional pseudo-elements extends the interface to ::backdrop, ::scroll-marker, and ::view-transition.\nCPU Performance API lets web applications determine the CPU performance of a user device.\nConnection Allowlists provides explicit network-level control over external endpoints using an HTTP response header.\nCSSPseudoElement support for ::backdrop, ::scroll-marker, and ::view-transition\nSupport for CSSPseudoElement in JavaScript, previously defined for ::after, ::before, and ::marker, extends to several new pseudo-elements in Chrome 152:\n::backdrop: Lets you handle interactions on modal dialog backdrops. For example, you can close a dialog when a user clicks the backdrop without interfering with clicks inside the dialog content, eliminating the need for complex intersection calculations.\n::scroll-marker: Enables interaction handling on scroll markers, such as collecting click statistics.\n::view-transition: Paves the way for geometry-aware view transitions and intercepting an active view transition mid-flight using coordinates of currently animating elements.\nFor more details on interacting with pseudo-elements in JavaScript, see the CSSPseudoElement documentation on MDN.\nCPU Performance API\nChrome 152 introduces the CPU Performance API, which lets web applications determine the CPU performance of a user device.\nWeb applications can use this tier information to adapt the user experience to the device capabilities—such as scaling graphical fidelity, adjusting heavy computations, or tailoring background tasks. This API can also be used in combination with the Compute Pressure API, which provides real-time data about CPU pressure and utilization.\nUsers can override the reported performance tier in Chrome browser Settings > Performance > Speed > Override CPU performance tier. Administrators can also control this behavior using the CpuPerformanceTierOverride enterprise policy.\nConnection Allowlists\nChrome 152 ships Connection Allowlists, a security mechanism that provides explicit control over external endpoints by restricting network connections initiated by a document or web worker.\nConnection Allowlists provide a direct method to address these risks by making the browser the gatekeeper of all network connections originating from your page. By including the Connection-Allowlist HTTP response header, a site specifies the exact URL patterns permitted for all network communication initiated by its context.\nTo learn more about configuration and reporting options, read Connection Allowlists origin trial.\nFurther reading\nThis covers only some key highlights. Check the following links for additional changes in Chrome 152:\nRelease notes for Chrome 152.\nChromeStatus.com updates for Chrome 152.\nChrome release calendar.\nSubscribe\nTo stay up to date, subscribe to the Chrome Developers YouTube channel, and you'll get an email notification whenever we launch a new video. Or follow us on Bluesky and LinkedIn for the latest updates.\n\nhttps://x.com/i/article/2092689904404697089","summary":"Chrome 152 introduces three critical capabilities for builders: CSSPseudoElement support enables fine-grained JavaScript control over modal backdrops and view transitions without complex workarounds, the CPU Performance API allows adaptive rendering based on device capabilities, and Connection Allowlists provide network-level security by restricting outbound connections via HTTP headers. These features matter because they address fundamental challenges in web development—simplifying complex UI interactions, optimizing performance across diverse hardware, and enforcing security policies at the browser level—directly improving both developer experience and application robustness.","author_name":"Chrome for Developers","author_handle":"chromiumdev","timestamp":"2026-08-26T19:08:47+00:00","comment_count":3,"like_count":61,"retweet_count":8,"submitted_at":"2026-08-26T19:15:24.75537+00:00","created_at":"2026-08-26T19:15:24.75537+00:00","seo_title":"Chrome 152 Adds JavaScript Control for Pseudo-Elements","bookmark_count":16,"image_url":"https://pbs.twimg.com/media/HQq6bYNWcAAUNi2.jpg","keywords":["web development","javascript","cybersecurity","software engineering","programming","web performance","api development"],"video_url":null,"video_thumbnail_url":null,"image_grid_urls":null,"profile_id":"eda3383a-d369-4f3c-b476-86e774ef0676","link_card_url":null,"link_card_title":null,"link_card_image_url":null,"link_card_domain":null,"is_pinned":false,"pinned_at":null,"content_type":"article","article_title":"New in Chrome 152","article_image_url":"https://pbs.twimg.com/media/HQq6bYNWcAAUNi2.jpg","slug":"new-in-chrome-152-2092690739130196410","quality_score":6,"noindex":false,"profiles":{"id":"eda3383a-d369-4f3c-b476-86e774ef0676","bio":null,"name":"Chrome for Developers","handle":"ChromiumDev","website":null,"location":null,"verified":true,"avatar_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/avatars/chromiumdev.jpg","banner_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/banners/chromiumdev.jpg","joined_date":null,"tweet_count":12076,"listed_count":5580,"follower_count":422295,"following_count":128}},{"id":"6569e182-0801-476d-9720-1b609c8f60d6","tweet_url":"https://x.com/mvanhorn/status/2092629365045559547","tweet_text":"📝 Every Grok Bot Hack I Know (Aug 2026)\n\n Three months ago I posted \"Every Agentic Engineering Hack I Know.\" It hit 1M views. Then I wrote up what people were using Grok Bot for, which was a ranking of strangers. This one is mine.\nGrok Bot is two weeks old and it already runs my inbox, my research, my meeting notes, my kids' calendar, my grocery cart, and as of this morning it tries to make my phone calls for me in Portuguese. I build last30days (59K stars), Printing Press (6.3K stars), and Agent Cookie, and all three are installed inside a bot that works while I sleep. These are my hacks.\nHACKS\nThe YOLO TL;DR Hack: paste this entire article into your bot and tell it to make a plan to set up everything in it, then work that plan one hack at a time. That is my whole stack, no reading required.\n1. Install Compound Engineering, and Make It Plan First\nRule number one in June, rule number one now. Agents perform better when they write a plan first. Not sometimes. Every time. A bot with a plan finishes the job. A bot without one cuts corners and stops early, then tells you it is done.\nSo the first thing I did to my Grok Bot was install Compound Engineering into it, globally, and then start every non-trivial job with a plan. ce-plan to think, ce-work to build. All day long.\nThe install is not in the plugin catalog, which stops most people. It should not. The skills are just folders, and Grok Bot can install them itself if you tell it to stop asking and go.\nHACKS\nPaste this into your bot:\n \n2. Give Your Bot an Inbox\nI have been giving my agents email addresses with AgentMail for a while now, well before Grok Bot existed. I liked it enough that I begged Adi to let me invest, and he let me. Then Grok Bot shipped and it turned out to be the best fit the thing has ever had.\nThe reason is the one Adi makes himself, and it is not \"email is convenient.\" The default move is to plug your bot into your own Gmail, which means every bot shares your inbox and sends as you. Your outreach bot annoys one prospect, your domain gets flagged, and now your personal mail stops delivering. One bot's mistake takes the whole fleet down. Give each bot its own real address and it gets its own sending reputation, its own threads, and its own blast radius.\n \nThat unlocks the thing chat cannot do. A bot with an inbox can sign itself up for a service and collect its own verification code without borrowing your identity. It can be CC'd on a thread and just participate, reading the history and replying in line like a person on the team.\nWhat I actually use it for is dumber and better. An email lands, I forward it to the bot, and I say \"add this to my wife's and my calendar.\" Done, from my phone, in the elevator. No app, no copy-paste, no context.\nThe one that earns its keep at home is the kids digest. Every morning it emails my wife and me a summary of what is on the kids' calendar that day. She does not use Claude Code. She does not use Grok Bot. She gets an email, which is a thing she already reads.\nOne setting, and it matters: it defaults to telling you every time it sends an email. Turn that off. Same rule as the Silence hack further down. If it worked, be quiet about it.\nHACKS\nGive each bot its own AgentMail inbox instead of plugging them all into your Gmail. Forward email to it instead of retyping the task.\nTell it explicitly to stop announcing every email it sends.\n3. Give It a Phone Number and Let It Make the Call\nI set this one up an hour ago. The technology works. The first restaurant I pointed it at hung up on it three times.\nI am on vacation, and the errand I kept hitting is the one that ends in a phone call. Call the restaurant and see if they can do eight at seven. Call the shop and ask if they actually have it in stock. Thirty second calls I did not want to make in a country where I do not speak the language, and the bot could not make them either, because it had no mouth.\nMartin Donadieu showed the shape:\n \nSo I wired up Twilio. It opens in English, and if whoever picks up answers in Portuguese it just keeps going in Portuguese, same voice, same call. The thing I was avoiding, calling a restaurant in a language I do not speak, turned out to be the exact thing it is best at.\nThe humans are another matter. Here is my first real call, a table for ten plus a high chair at a farm restaurant in the Algarve:\n \nPedro picked up, they switched to Portuguese, and he hung up at fourteen seconds. Called back, hung up again. Third try with the full ask in the first sentence, hung up in nine seconds. Then my bot stopped on its own: \"Stopped after three tries so we don't look like spam.\"\nSo I called the restaurant myself, like an animal. A different woman answered and booked the table in about forty seconds. She told me Pedro did not like the robot and would not talk to it.\nThat is the honest state of this hack. The call connects, the language switching is genuinely good, and the bot even has better manners than most people would, but a real person on the other end can simply decide they are not doing this today. Small businesses in Portugal are not expecting a robot to call about lunch. Do not put anything time-critical on it yet, and if it matters, be ready to dial it yourself.\nThe setup, at least, is the hack from section 1 all over again. I did not read a Twilio tutorial. I told the bot what I wanted and it wrote its own install prompt, and that prompt is below verbatim. Bland is the other option if you would rather not assemble it.\nStill on my list: WhatsApp, because outside the US that is where the reservation actually happens, and because a message does not hang up on you.\nHACKS\nPaste this, and have your own live Twilio account and number ready. Expect some humans to hang up:\n \n4. Turn a Bot Into last30days\nThis is the one I use most. Not last30days as a tool I run, but a bot that IS last30days, so research is a text message.\nRename the bot to last30days. Then when a topic comes up, on a walk, in a meeting, in the car, you text it and a full sweep across Reddit, X, YouTube, TikTok, Hacker News, Polymarket, GitHub, and the web comes back while you keep going.\nHonest note: X as a source is a little wonky right now, cookies and rate limits both. I have faith there is a real fix landing soon, and the other eight sources carry the sweep in the meantime.\nHACKS\nPaste this:\n \nThen cron it. Weekly sweeps on your business, your competitors, and yourself.\n5. One Front Door, and It Delegates\nYou talk to one bot. It talks to the others. This is the structural decision that separates the fleets people keep from the fleets people quietly stop opening.\nThe clearest version of it is Nate Herk's, in A Week of Grok Bot Lessons in 10 Mins on YouTube, 23,509 views. His front door is a bot named Klaus, and the rule is that Klaus is the only bot he talks to. The instruction that makes it work is a delegation check, run before Klaus does anything himself:\nBefore doing any task, check whether another Grok bot owns it and delegate first.\nNate Herk, YouTube, 23,509 views\nMatt Shumer ran the experiment most of us assumed would fail. He made a researcher bot and a writer bot, then made a Chief of Staff and asked it to get the two of them working together, fully expecting it to fall apart. It worked out of the box.\n \nPrajwal says the same thing from the other end: you put your bots in one group chat, appoint a manager, and stop being the router between your own tools.\nOne correction while you plan the fleet, because almost every walkthrough video gets it wrong. Your bots do not each get their own computer. The docs are explicit: they share one persistent cloud computer, each with its own screen. Several can drive the browser at once, but one bot runs one computer-use task at a time. You parallelize by splitting work across bots, never by piling it on your favorite one.\nHACKS\nPaste into your front-door bot:\n \n6. Hire a Bot Advisor to Build Your Other Bots\nThis is my favorite hack in the whole article and I did not invent it. It is the Bot Advisor on botdirectory.ai, the second most copied prompt on the site, inspired by Peter Yang.\nA bot whose only job is to create and manage your other bots.\nThe reason it matters: the thing that kills people in week two is not bot capability, it is bot sprawl. Eleven bots, four of which you forgot exist, two of which are noisy, and none of which you can remember the instructions for. The Bot Advisor is the fix. You describe something you want automated, it interviews you, it drafts the instructions and the schedule and the connections, and it creates the thing. When a bot gets noisy it tightens that bot's instructions. It keeps the roster.\nGive botdirectory.ai a look before you write a single instruction block from scratch. It is the copy-paste layer for this whole product.\nHACKS\nPaste this:\n \n7. Permanent Named Roles, and Fire the Ones That Do Not Work\nThe best story of the launch is not from an engineer. Jon ONeill runs a plumbing company. He spent a week getting Claude to about 80 percent on work-order intake across Gmail, Slack, ServiceTitan, and customer portals, with no end in sight. Then:\n \nNow a bot reviews five or six time-sensitive work orders a day, finds calendar slots, confirms the customer, and books an actual plumber. Ten minutes, start to finish.\nBut the part worth stealing is the failure. His first bot was Dana Dispatcher, an end-to-end dispatcher, and he fired her. Too slow, too many mistakes. So he split the job by software instead: Gary listens for new work orders in Gmail, Steve checks the ServiceTitan board for capacity, Quinn texts customers and sends the accepts back to Steve, Sarah posts to Slack for the human staff, Marc manages the specialists, Atlas is the right hand. He took the blame for the bad job description, not the bot.\nThey still make mistakes from time to time, but what I'm finding is they're not any worse than mistakes we've had real life humans make when we're training or onboarding them. So we teach, train, correct.\n@HouseHackerJon, X\nThe names are not a gimmick either. Alex Finn, in a 54,789-view walkthrough, points at what most people actually do:\na majority of people I see using Grok Bot just like make the names of each of their bot like content, email, coding\nAlex Finn, YouTube, 54,789 views\nMatthew Berman, who has used it daily since launch, names why that costs you something:\nThe UI is built to give every bot the feeling of a persistent character rather than a thread that you can just throw away.\nMatthew Berman, YouTube, 42,038 views\nA thread called \"email\" is disposable. GranolaBoy is not. Look back at my roster: the ones with human names are the ones that survived, and the one still called New Bot has never done a thing. The naming is a commitment device pointed at you, and you delegate more readily to something you think of as a someone.\nHACKS\nOne job per bot, named like a person, split by the software it lives in. If you cannot name what a bot does every week, delete it.\n8. Write the Instructions in Three Layers\nStop writing one paragraph in the description field. Write three stacked sections. It is the highest-leverage fifteen minutes you can spend on this product and almost nobody does it.\nThe base layer is where you put the things you are tired of repeating. The role layer is where you put the standard of proof, which is the part that actually changes output quality. A researcher bot with \"cite the source URL for every claim\" in its role layer stops handing you confident summaries of nothing.\nWorth knowing: on iOS the field has been renamed from Description to Instructions while desktop still says Description. That is one user's observation, not an announcement, but Instructions is the better name and it tells you what belongs there.\nHACKS\nPaste and fill:\n \n9. Silence Is the Hack\nJohn Viklund posted the version of this that made it click for me. He snaps a photo of a tag in a store, the bot pulls brand and SKU and size, saves a price cap, then scans daily and pings him only on a real hit.\n \nHis line: that silence is the actual hack. He is right, and it is the thing that separates a routine you keep from a routine you mute in four days.\nPrajwal published the two best worked examples of the rule. His wave radar: \"Ping me ONLY for things I could post about in the next 6 hours. No engagement bait, no reposts of old news.\" His night moderator: \"If you're less than sure, don't answer. Add it to a list for me instead.\"\nMy version is research. I have weekly last30days runs cron'd on my own business, on my competitors, and on myself. They run Sunday night. Most weeks they say nothing, because nothing happened worth saying. That is the point.\nThe other half of silence is picking the cadence from the work rather than from enthusiasm. A fifteen minute patrol is for people whose timing is the product. Always-on is always-spending.\nHACKS\nAnswer these five before you create any routine:\n \n10. Screenshot It and Text It to Your Chief of Staff\nThis is the simplest thing in the article and I use it more than anything else here.\nSomeone sent me a LinkedIn message asking me to speak at a conference. I did not open a calendar. I did not copy the date out, or retype the venue, or forward it anywhere. I screenshotted the message, sent the image to my Chief of Staff, and said \"add to my cal and my wife's cal.\" Both calendars, done, from the LinkedIn app.\nThe reason it works is that the screenshot is the input. You are not the parser. Every date, name, address, and time in that image is already there, and the bot reading it is the same thing that is going to write the calendar entry. The second you find yourself retyping something off a screen into a prompt, you have done the bot's job for it.\nIt works on anything you can capture. A text message with a birthday party address. A school newsletter. A flight confirmation. A photo of a flyer on a door. A screenshot of an error. If you can see it, you can hand it over.\nHACKS\nScreenshot anything with a date, an address, or a task in it, send the image to your front-door bot, and say what you want done with it in plain words. Never retype what is already on the screen.\n11. Never Let Your Bot Be Logged Out\nEvery hack in this article dies the moment your bot gets logged out of something. That is the real failure mode of cloud agents, and it is boring, which is why nobody writes about it.\nAgent Cookie fixes it. It securely syncs your Chrome cookies from your Mac to your Grok Bot in the cloud over Tailscale, so the bot is logged into what you are logged into, without you pasting a password anywhere.\n \nHere is the receipt I got this morning. I had not touched Instacart on Grok Bot in a week. New computer since then. New address, too. I texted it \"can you add greens lettuce to costco\" and it came back first try, no re-auth, no login screen:\n \nThe piece that makes that survive is PR 119, which resolves Tailscale hostnames at sync time and auto-rebinds stale sink IPs. A new bot comes online, it lands on the same Tailscale net without me doing anything. I run it every fifteen minutes so the link never goes stale.\nHACKS\nInstall Agent Cookie and sync your Mac's Chrome session to your bot over Tailscale.\nRun the sync every 15 minutes on a routine so a new bot inherits the same net.\n12. Printing Press CLIs, Inside the Bot\nComputer use will drive any website, and that is genuinely the magic. John Ennis put it well: he now hands Grok Bot the complicated services he never bothered to learn and calls it a universal interface.\nBut for the ten services I touch every single day, I do not want a browser session and a screenshot loop. I want a command. That is Printing Press: agent-native CLIs that wrap real services so the bot just does the errand, fast and deterministically.\nThese are the ones living in my bot right now. ESPN watches a game and pings me when it gets close. Instacart puts lettuce in the Costco cart. flight-goat and ticketdata handle travel and seats. arxiv, digg, and techmeme feed research. twilio and x-twitter handle messages and posts.\nComputer use is the fallback for everything else. The CLI is what you reach for on the site you live in.\nHACKS\nInstall the fleet:\n \nBrowse the rest at printingpress.dev, then print your own for the service you use all day.\n13. Give It Peekaboo and It Can Drive Your Mac\nThis one is Ben Lang's, from inside the company, and it is one sentence:\n \nHere is why it matters, because the sentence undersells it.\nYour bot's computer is a Linux box in the cloud with a browser on it. That is the whole world it can touch. Anything with a web app, it can do. Anything that only exists as a native Mac app, it cannot see at all. Messages, Photos, Finder, Preview, Xcode, the desktop Slack client, Final Cut, and every crusty desktop app your work depends on that never shipped an API. All invisible.\nPeekaboo, from Peter Steinberger, is a Mac tool that sees the screen and does the clicks. It screenshots any app or the whole system, and it reads the accessibility tree, so it does not just get a picture, it gets a structured map of every button and field in an app with an ID for each one. Then it clicks them and types into them, targeted at a specific window. The loop is: look at the screen, pick an element, act on it, look again.\nPoint your bot at that and the cloud box stops being the boundary. It can pull a photo out of Photos, rename fifty files in Finder, read a text thread in Messages and do something about it, or click through a desktop app that has no API and never will. The thing that has been stopping your bot is not intelligence. It is reach.\nOne thing to keep in mind: this points an agent at your actual machine rather than a disposable cloud box, so everything in the security hack below applies harder here.\nHACKS\nAsk your bot: \"set up Peekaboo on my Mac, then list the native apps you can now see and control that you could not before.\"\nStart it on something reversible. Renaming files in a folder, not sending things from Messages.\n14. Install Claude Code On Its Computer\nSame principle as Peekaboo, one level weirder. The bot's cloud computer is a real computer, which means you can log into Claude Code on it.\nAusten Allred posted the moment somebody did, and his two words are the correct reaction:\n \nAsh Tilawat logged into Claude Code on his Grok Bot's computer, then made a PM bot, a Dev bot, and a QA bot, and showed them how to clone themselves. Austen's caption: \"And so it begins.\" 40K views and 213 likes in a few hours, mostly from people realizing what that sentence means.\nThat is a real change in what this thing is. Grok Bot is the always-on lane, Claude Code is the deep-work lane, and now the always-on one can drive the deep-work one while you are asleep. That is a genuinely different machine than the one xAI shipped two weeks ago.\nI have a Claude Code Bot in my roster and it is still parked in the experiment column, and this is exactly why. Read the last section before you teach anything to clone itself. Bots that spawn bots with nobody watching is the precise shape that multiplied errors 17 times over in the only real comparison anyone has run.\nHACKS\nAsk your bot: \"log into Claude Code on your computer and tell me what you can do now that you could not do before.\"\nDo not let it clone itself. Give it one job, watch it beat that job, then decide.\n15. Every Meeting, One Question Away\nI have a bot called GranolaBoy. It was supposed to be Granola Bot. I typo'd it and I am keeping it, because it's funny.\nIts whole job is my Granola meeting notes. Every conversation I have had, searchable by asking. What did that candidate say about pricing. Who was the person who mentioned the warehouse thing three weeks ago. What did I promise in the Tuesday call. It answers, with the meeting and the moment.\nThis is the compounding one. The June version of this hack was \"put the raw transcript in your LLM, do not summarize first.\" That still holds. But the upgrade is that the whole archive is now a bot I text, and every meeting I have makes the next answer better.\nHACKS\nPoint a dedicated bot at your meeting notes and let it answer across all of them, not one at a time. Never summarize before you hand it over.\n16. Teach by Recording, and Respect the Ten Minute Cap\nRecording a workflow instead of describing it is the feature people keep independently rediscovering. Dillon Loomis has the version worth studying. He screen-recorded himself cleaning a messy desktop, talking through why he sorted things the way he did:\n \nHe had planted a trick question in the recording. It answered correctly.\nNow the constraint that is in the docs and in almost none of the videos: the recording caps at ten minutes. Your real workflow is forty minutes long, so recording it in one pass fails, and people read that failure as the product not working.\nHACKS\nSplit any long workflow into named stages under ten minutes and record each one:\n \n17. Files Are the Memory Bus, and Your Bots Are Not a Security Boundary\nThe most misunderstood thing about this product, and it explains a failure people keep blaming on the model. Your bots do not share conversations. They do share a filesystem.\nSo when you tell your research bot something important and your writing bot has never heard of it, that is not forgetfulness. Those are two separate memories. The three things that actually cross the boundary are files, group messages, and explicit handoffs. Which means the fix is a convention, not a prompt.\nThe harder half: all of your bots use the same persistent cloud computer, sharing files, browser sessions, and app logins. The docs say it twice, plainly. Do not use separate bots as a security boundary. Everything you connect, every bot can reach. Scope by what you log in to, never by which chat window you are typing in. Prajwal's rule for week one is the right one: read-only on public stuff first, credentials later.\nHACKS\nSet the convention, then put the last line in every bot's base instructions:\n \n18. Draft-Then-Approve, Then Point One Bot at Money\nEvery setup I trust has a human gate at the same place: money or send. And every person running one found that boundary before they needed it rather than after.\nThen, once the gate exists, point one bot at something with a number attached. Darian Shirazi has the cleanest receipt of the launch. He asked his bot to pay its own monthly fee, it went through his email hunting lost money, emailed five merchants that had never refunded his returns, and it has now made back more than the subscription costs.\n \nAlex Finn's is bigger and the same shape: an inbound sponsorship email, answered and negotiated by a bot, closed at $10K.\nBoth are narrow and inbound. The anti-pattern is the one that gets accounts banned: do not tell a sales bot to pull every contact in the state and message all of them. A bot has no tedium and no embarrassment, so it will do exactly that, at a volume that reads as spam to every system it touches.\nHACKS\nSet narrow rules, not vague ones. \"Be careful with money\" is not enforceable.\n \n19. Steal From the Lists, and From the Replies\nYou do not have to invent any of this. Four lists are carrying the whole community right now, and they are all free.\nEric Zakariasson's 100 use cases is the deepest. Ben Lang's internal most-loved list is the one from inside the company. Miles Deutscher's 25 ways is the most copy-pasteable. Peter Yang's five-bot tutorial is the best starting point if you have zero bots today. And botdirectory.ai is the prompt layer under all of it. One honest caveat on the directory: rank there is copy count, not community-tested quality. Read the prompt before you run it.\nThen there are the replies, which have been better than the lists. Two I keep going back to.\nKun Chen runs open source projects with 24K stars between them and was drowning in issues and PRs. He wrote a VISION.md for every repo, built a software factory with Grok Bot plus Cursor cloud agents, and the queue finally started moving:\n \nThen he did the generous thing and packaged the whole setup so you can run it too. It is called Grok Ship, and the pitch on the repo is four words: turn your Grok Bot into a software factory. You do not clone it or read it. You tell a bot to go get it:\n \nThen, per repo: \"set up a crewmate to auto triage this repo.\" That is a maintainer's entire triage problem handed off in two sentences.\nAnd Trevin, who I build Printing Press with, has an OSS bot triaging and labeling issues and PRs in public repos, merging them against criteria he set, and pinging him only when something is ambiguous or strategically important. He also has a DJ bot on Spotify that builds a playlist sized to twice his drive time whenever his calendar has a drive over ten minutes:\n \nHACKS\nRead Peter Yang's tutorial first, then Miles's 25, then Eric's 100. Copy the closest prompt from botdirectory.ai and edit it. Do not write from scratch.\nMaintainers, paste this: \"setup Grok Ship for me. follow GROK_SHIP.md in this github repo: kunchenguid/grok-ship\"\n20. Run It From Your Phone, and Talk to It\nThe mobile app is where this product is actually different, and the voice UI on it is good. Same argument I made in June: voice into an LLM works because the listener understands context. It guesses what the mic got wrong. You can mumble, trail off, restart the sentence, and the bot still gets it.\nThe difference now is that voice on my phone is not dictating into a session on my Mac. It is talking to something that is already running, that has its own computer, and that will still be working after I put the phone down. I sent the Instacart request from the kitchen. I sent this article's research request from a chair.\nPrajwal has the receipt for what that adds up to:\nI ran all 5 of my businesses from my phone today... Yesterday someone asked me how many people work for me. I said none.\n@PrajwalTomar_, X\nHACKS\nUse the voice button in the mobile app, not typing. Lazy sentences are fine, the bot fills the gaps.\n21. The Honest Part: The Meter, the 17x, and the Counting Test\nThree things I would want told to me before I set any of this up.\nThe price moved and most guides still have it wrong. As of August 21 it is on Cursor Pro+ at $60 a month, SuperGrok Plus at $100, Cursor Ultra at $200, SuperGrok Heavy at $300, and Cursor Teams Standard at $40 a seat, with a limited free trial:\n \nWhat the price headline hides is that the meter above each plan's weekly allowance did not move, and under real fleet work it moves fast. The quota complaints are specific and they are from people who like the product. Do not start on the $300 tier. Start on the cheap one and watch what a fifteen minute patrol actually costs you before you make three more of them.\nSecond, unsupervised crews are not free. Prajwal ran the comparison and found crews with no supervisor multiplied their own errors up to 17 times the solo rate. That is why every bot in his setup runs solo, one job, one output. A wrong step lands on his desk instead of infecting three other bots.\nThird, the counting test. Count how many of your bots actually own an outcome. Not how many exist, not how many have clever names. How many own a thing that would visibly not happen if they stopped.\nI ran it on myself while writing this. Here is every bot I have, including the half-built ones and the one still named New Bot:\n \nFifteen bots, eight that own something. GrokLawyer drafts boring legal language maybe twice a month. Trip Planning Bot solved one drive and has sat there since. New Bot is named New Bot. I am showing you all of it because every article about this product shows you a tidy fleet of five, and nobody's actually looks like that.\nThat is the honest shape of a two-week-old fleet, and it is fine, as long as you know which half is which. The failure is not having experiments. The failure is telling yourself the experiments are staff.\nHACKS\nStart on the $60 tier and check the meter before you scale the fleet.\nRun bots solo until one of them beats the job.\nCount the bots that own an outcome. Delete the ones that do not.\nCopy this whole article, paste it into your bot, and tell it to set up everything it can.\n\nhttps://x.com/i/article/2092207568907071488","summary":"# Why This Matters\n\nThis post establishes practical, production-ready patterns for deploying autonomous agents at scale, transforming Grok Bot from a novelty into a reliable infrastructure layer that handles email triage, calendar management, research, and business operations while the user sleeps. For builders and founders, the 21 documented hacks—from implementing planning-first architecture to establishing single points of delegation to preventing unsupervised bot error multiplication by 17x—provide immediately deployable blueprints that compress months of experimentation into copy-paste configurations, fundamentally changing the economics of knowledge work automation.","author_name":"Matt Van Horn","author_handle":"mvanhorn","timestamp":"2026-08-26T15:04:54+00:00","comment_count":11,"like_count":215,"retweet_count":15,"submitted_at":"2026-08-26T16:05:24.016023+00:00","created_at":"2026-08-26T16:05:24.016023+00:00","seo_title":"Grok Bot Hacks: How to Build a Fleet of Agents That Actually Work","bookmark_count":722,"image_url":"https://pbs.twimg.com/media/HQpjG3zXsAA8F8M.png","keywords":["ai agents","automation workflows","prompt engineering","agentic engineering","bot productivity","cloud computing","software automation","task delegation"],"video_url":null,"video_thumbnail_url":null,"image_grid_urls":null,"profile_id":"25aa531f-4a68-4679-875a-64a725ba6736","link_card_url":null,"link_card_title":null,"link_card_image_url":null,"link_card_domain":null,"is_pinned":false,"pinned_at":null,"content_type":"article","article_title":"Every Grok Bot Hack I Know (Aug 2026)","article_image_url":"https://pbs.twimg.com/media/HQpjG3zXsAA8F8M.png","slug":"every-grok-bot-hack-i-know-aug-2026-2092629365045559547","quality_score":8,"noindex":false,"profiles":{"id":"25aa531f-4a68-4679-875a-64a725ba6736","bio":"Co-founded June (“self-driving oven,” acquired by @webergrills) & co that became @Lyft. Building again, more soon. OS: @slashlast30days 47k★ @ppressdev 5.4k★","name":"Matt Van Horn","handle":"mvanhorn","website":"http://github.com/mvanhorn","location":"Seattle, WA","verified":false,"avatar_url":"https://dcuhjhnlzlpnshikrfwz.supabase.co/storage/v1/object/public/tweet-media/avatars/mvanhorn.jpg","banner_url":"https://pbs.twimg.com/profile_banners/6238012/1348811035","joined_date":"2007-05-22T16:48:48+00:00","tweet_count":16596,"listed_count":748,"follower_count":38702,"following_count":4883}},{"id":"9917043e-6a8d-4ecf-acb5-4ad8a8a00103","tweet_url":"https://x.com/github/status/2092386542304342311","tweet_text":"📝 35 free and open source games to celebrate 35 years of Linux\n\nBy @leereilly\nLinux turns 35 today, so we’re taking the opportunity to highlight 35 of our favorite free and open source (or “source available”) Linux games, their communities, and their stories!\n \nIf you're into video games, then you're bound to find something you like below. Oh, and some of the games work on Windows and macOS too, so there should be something for (almost) everyone.\nWhat if you could play a kart racer, a space sim, a city builder, a roguelike, and an RTS... and never spend a dime?\nYou can.\nHere are 10 free and open source Linux games worth checking out. 🎮 🐧\n \nWant to build an empire, conquer the galaxy, or lose 40 hours to a dungeon?\nThere’s an open source game for that.\n➡️ Here are 10 more free Linux games you might not know about. \n \n \nWhat do you get when you combine Command & Conquer, DDR, Minecraft, tower defense, factory building, and open source? This.\nThe final 10 (actually 11, due to a classic off-by-one error) free and open source Linux games in the series.\n\n\nWe said 35...\nThen we found a few more games too good to leave out.\n#32: Beyond All Reason\nFirst up: Beyond All Reason, a massive open source RTS where the battles can get seriously out of hand... especially if you concentrate on building advanced fusion reactors or forget to build a fighter wall...\n \n \n#33: Space Station 14\n\nWhat started as an open source remake of Space Station 13 almost died. Twice.\nTen years later, Space Station 14 has thousands of concurrent players, hundreds of contributors, and is on Steam.\nA great example of what an open source game community can build together.\n \n \n#34: OpenRCT2\n\nRollerCoaster Tycoon 2, but open source, modernized, multiplayer, and still being actively developed.\nOpenRCT2 lets you build the amusement park of your dreams. Or the amusement park of your nightmares.\n \n \n#35: Hedgewars\n\nRemember Worms?\nNow imagine an open source version with hedgehogs, ridiculous weapons, destructible landscapes, and multiplayer battles.\nThat’s Hedgewars.\n \n \nNow grab a pumpkin spice latte and go check them out on your next coffee break!\n\nhttps://x.com/i/article/2092335686401810432","summary":"This post matters because it demonstrates that high-quality, feature-complete games can be built and sustained through open source collaboration, providing builders and founders with proven examples of community-driven development models and showing AI practitioners potential training data sources and game environments for machine learning projects. For practitioners evaluating Linux ecosystems and open source viability, the diversity of genres—from RTSs to city builders to multiplayer platforms like Space Station 14—proves that open source can compete with commercial alternatives while building engaged communities across multiple platforms.","author_name":"GitHub","author_handle":"github","timestamp":"2026-08-25T23:00:01+00:00","comment_count":25,"like_count":2152,"retweet_count":282,"submitted_at":"2026-08-26T20:15:46.658684+00:00","created_at":"2026-08-26T20:15:46.658684+00:00","seo_title":"35 Free and Open Source Games for Linux","bookmark_count":466,"image_url":null,"keywords":["open source gaming","linux games","free software","gaming communities","open source development","game design","software engineering"],"video_url":null,"video_thumbnail_url":null,"image_grid_urls":null,"profile_id":"835d6cc6-dd91-48b2-95ac-d5dfa1bcdada","link_card_url":null,"link_card_title":null,"link_card_image_url":null,"link_card_domain":null,"is_pinned":false,"pinned_at":null,"content_type":"article","article_title":"35 free and open source games to celebrate 35 years of Linux","article_image_url":null,"slug":"35-free-and-open-source-games-to-celebrate-35-years-of-linux-2092386542304342311","quality_score":4,"noindex":false,"profiles":{"id":"835d6cc6-dd91-48b2-95ac-d5dfa1bcdada","bio":null,"name":"GitHub","handle":"github","website":null,"location":null,"verified":false,"avatar_url":null,"banner_url":null,"joined_date":null,"tweet_count":3,"listed_count":0,"follower_count":0,"following_count":0}}],"total":395,"limit":20,"offset":0,"nextCursor":"eyJ2ZXJzaW9uIjoxLCJzY29wZSI6ImFydGljbGVzIiwidGltZXN0YW1wIjoiMjAyNi0wOC0yNVQyMzowMDowMS4wMDBaIiwiaWQiOiI5OTE3MDQzZS02YThkLTRlY2YtYWNiNS00YWQ4YThhMDAxMDMifQ","hasMore":true}