claire vo

I used to have to coach underwater managers to take time out of their day and focus on recruiting, now I have to tell people to pause and train their agents well

Not at all overkill: in the last 24 hours I built a electron app to backup/sync my @openclaw knowledge file .md file system between my device and hers, track agent tasks, assign between agents, and keep a log of the briefings she pushes to me regularly. ofc I protect her soul 💕

📝 You've been kicked out of the arena, you just don't know it yet If AI adoption had 7 stages of grief, almost all of you would be in denial. No matter how many AI memos your CEO sends, the amount of Claude that's being Coded, the chatbots in app and the evals in data--I'm here to tell you: you're not competing. In fact, you probably can't anymore. And you won't notice until it's too late. A going concern = the walking dead I look around and see a lot of businesses that would have, 3 years ago, been pretty enviable. $100M+ in revenue, solid customer base, mature executives, and a good enough team. On the surface, a few of them are trying to meet the moment. Whether by choice or by force, AI has worked it's way into little parts of the business and the product. They have some internal AI power users. They're slapping on chatbots and MCPs to their product suite. They have a natural-language interface for their product named after a girl or a gem or a spaceship. They start to say things internally like we need to be "cursor for X" or "claude code for y" or (perhaps worse) "we have a proprietary data moat they'll all need," and in the meantime the pot keeps boiling. While they try on the bows and ribbons of an AI-native team, they ignore the fact that their bones are old and the company has calcified. For the most part: sales still sells the same and marketing is still talking about channels and CAC and product says "prioritize" and eng says "capacity" and the board is endlessly asking either about Q1 perf and Q2 projections or the ever-elusive "increase in product velocity." Some, but not all, engineers are coding with AI > 50% of the time. Some, but not all, designers have access to the main repo, will gently ask for a PR. Some, but not all, the roles that could have been replaced have been. There are still so many decks, so many meetings, so many things that are pushed to later or when we have headcount or after we wrap this quarter. They feel busy and look at their revenue and their customers and the movement (note: not the speed) of the team and they think "I'm in the arena!" Your peers aren't your competition anymore The reason it's so easy to convince yourself and an entire company that you are totally in the game is you think your competition is who they've always been and will act about the same as they always have. Because guess what? You're who you've always been and you'll act the same as you always have. You'll be in the same deals. Have the same marketing tactics. You'll ship equal proportions of new product and make noise about them in just about the same way. You'll call on and hire the same class of VPs, the engineers with the same YOE and backgrounds, the same PMs with "domain expertise." You'll listen to the same podcasts and forward the same tweets. You'll have the same ideas for how AI is going to transform your business, and you'll pitch the same mini-pivots to the board. You'll all say you'll do this without alienating the core. Without giving up what you've built so far, that you're so good at, that no one else could possibly do. But this is not your competition anymore. There are startups that are doing everything you claim to be doing with AI, but without thinking about or working too hard at it. They naturally reach for the right tools and have no processes they need to circumvent to get access to them. While you are debating what % of R&D should be allocated to AI initiatives, they have built the product idea you have sitting on a slide, called on your customers, and established their wedge. You probably won't know until renewal time. You probably won't appreciate it until your best talent starts to leave & work for them. You won't know it till you're cooked. Everything > 1 day: ngmi I have a simple heuristic for identifying who has been kicked out of the arena and it's this: They can't do anything the same day. It doesn't matter what it is, but I'll give you a list of things I've seen companies be unable to do in a day: Fix bugs Launch landing pages Upgrade to a new model Get a PM on a call with a customer Get access to a new AI tool Turn around negotiations on executive comp Before AI, this was a less useful heuristic because many of these took scare expertise (engineering) and therefore time. But now we have the clarifying light of AI: if you can move fast and yet you don't--it's a you problem, babe. Startups: don't underestimate the level at which the big guys won't compete with you I'm not sure if it will be healthy to give startups more swagger at this point (from what I see, most have it in spades), but I still think some teams completely underestimate the level at which incumbents have taken themselves completely out of lucrative games. Markets that previously felt like winner-has-all are wide open. For two reasons: The fundamentals of what product is is changing so much, there's ability and appetite for new things in old markets. These companies cannot get out of their own way if their lives ARR depended on it (it does) As a startup, I'd take bigger bets, be more aggressive in the market, and not only take market share from your friendly neighborhood BigCo, but I'd steal their best employees, too. I'd simply DM the ones I wanted about the cost of staying and see who replies. I'd code up a ripoff of 80% of their core product, and launch it on X for 20% the cost. I'd be obviously Having More Fun™️ online. I'd play a game they're too busy in QBRs to notice. How to avoid fate There are a few, a very very few, who are the exception here. They are the ones with the CEO pushing PRs, the VPE DM-ing with users (not buyers! USERS!), the unhinged kid with little oversight on marketing. They will rip out entire functions, they will fire executives, and they will halt everything in service of not missing a generational technology shift. They'll say "F the core" and buy who they need to buy to rebuild the company. They'll (yes) bring the team back to the office. Or they'll rearchitect the entire company in .md one weekend, open Claude Code during the all hands and say: this is the new office. They might, just might, have the guts and the capital to make it. But probably not. But I can't blame you for wanting to try, so if I were at one of these companies this is what I would do: Everyone up-skilled immediately, stop the presses, show up in 2 weeks and show me how you've automated 80% of your job. If you've done this--congrats! You get a new more exciting job. If you haven't... what are we doing here? Get legal + security + finance to stop being annoying about trying new tools. Let your best engineers loose on making your repo agent-friendly. Resistant execs out the door. Resistant engineers out the door faster. More aggressive than comfortable M&A. You probably can't build what you want. You might be able to buy it. You should know every one and everything worth acquiring in your market, and try to get to them early. Rebuild your product from scratch. Just try. If it takes < 2 week, ask yourself what that means from your company. Hire executives sparingly, and carefully. Same day SLA for anything (bugs, polish, copy) that can be done well by AI. Unabashedly learn from whoever you can. Beginners mindset, especially if you're a CEO or exec I'm happy to help you try. But it'll be no use if you don't realize the arena has packed up and left town without you. So take a look around. Ask yourself: where am I? And more importantly: where am I not? http://x.com/i/article/20238728098067456…

Multi-modal AI is not just a party trick, it's a real unlock for making life (and work) more accessible. That's what Joe, principal software engineer at @babylist, is doing with Claude Code. After losing most of his vision in college to a rare genetic disorder, Joe had to figure out ways to do work on a screen with limited central vision. Then came AI, which was not only a huge unlock for his productivity as an engineer, but also helped him build custom tools to make software more accessible. In this ep you'll see Joe: - build slack extensions to "read" images - use claude code + vs code to develop a suite of useful chrome extensions - share how he develops claude skills to extend repeating patterns You'll also hear the most heartwarming use of Meta AI glasses yet 😎🫶 This is an inspiring ep, both for the 10x engineers out there as well as anyone thinking about the intersection of AI + accessibility. A big thanks to our wonderful sponsor @tines_hq - start building intelligent workflows today Watch now: https://youtube.com/watch?v=sibufEEhH6A

Gonna revive Polly the @openclaw this weekend and see what’s new 🦞 Ultimate goal is she can do all human-facing things for me so I can retreat into a hoodie, become one with the tokens, and code in silence. But first I gotta make her be able to tell time.

AI has raised the bar on the basics, and is creating insane competition for talent. I get asked all the time who is a great CTO/CPO/vp prod/PM/staff eng all the time. And usually my response is: “there is no one good you can get.” Because it’s not “can someone take this seat” anymore. It’s literally: is this a top 5% person I’ve seen do a job like this, are they AI pilled, are they not a founder or snatched by a research lab and can you compete with their comp. It isn’t: here’s a solid VPP w 15 yoe who has been in your vertical for while, they seem pretty good. It’s: this person is worth their weight in Claude Code. This person adds to the system. This person *cooks*. Even when I share names, GUARANTEED within weeks they announce YC or a rad oai job or they secretly went to a superintelligence lab. (Maybe this is a me problem, and I’ve just worked exclusively with super talented people.) But there just aren’t enough, truly, out there to fill the demand. Tech twitter is not LinkedIn. There are still more jobs than people w the skills to fill them. Which means there is TONS of arbitrage right now in a) learning and building a AI portfolio and b) sharing in public/networking. Good isn’t good enough. Your track record means nothing now if you haven’t met the moment. Go learn, go build, and the hmu - i probably know someone who wants to hire you!

I've been testing the new OpenAI Codex app for a week. For those catching up, it's a GUI on top of the codex harness focused on the annoying problem of agentic local development: git (worktrees, branches, commits, PRs) and working across local + the cloud. # What makes Codex different? ## Very focused Well first, it's not a TUI. It's not a chatbot. It's not an IDE. It's... something else. Codex strips away everything but the primitives it's focused on, which are: • Projects (basically, a repo) • Threads (chat) • Git things (worktrees, branches, diffs, commits, PRs) • Automations • and Skills I guess they realized everyone is yelling at their coding agent "SORRY WRONG BRANCH" and packaged up some of the core workflows that AI-powered engineers use into a graphical interface. ## Skills - now with icons! Codex also addresses my primary gripe with skills, which is they are very powerful but seem less cool when presented as markdown + code files (or worse, a .zip) In the Codex app, skills are they're own page, and installable with a button. It's obvious in the UI when you've tagged in a skill to an agent, similar to how you'd @ mention a file or branch in Cursor. The library of skills is a mix of technical and non technical use case, which point to the discussion I always have with these platforms which is: "who are we building for"? To be honest, I'm still partial to a npx skills add from skills.sh, and I suspect most software engineers are, too. ## Automations Now this, I love. Because codex can work in the cloud, the ability to automate work (yes, while you sleep) and using nice UI to set it up is awesome. A lot of these automations are things I know many developers are already doing via github actions or some other agent, and it's nice to see this get first class treatment, instead of being a tagged on hacked use case. # My coding experience with Codex I tried a few things with Codex including • Restructure of a components folder • Reviewing complex work executed by Claude Code for quality • Drafting a technical spec for a personal agent • Checking the codebase for a specific inconsistency OpenAI has great coding models. Everyone has great coding models! We're spoiled for choice. I love GPT-5.2-Codex for backend things. I use it to address all my Bugbot feedback. It's a great model. Unfortunately, my experience with the Codex harness blunts my appreciation for the power of the model. You all know I'm a @cursor_ai daily driver, and it's clear that they've filed away a lot of the sharp edges of using a lot of different models in their app. Codex isn't as polished here, which meant coding with Codex felt so. dang. slow. Not because the model was particularly slow (though there was some of that), but because the harness kept getting in the way of me going yolo. "Always run in this session" didn't always run commands within the session. The app was waiting on me.... a lot. The integration between terminal, browser, etc. was behind other platforms and it made my overall developer experience sub par. Also, this is a personal thing, but she's so dry. I love an efficient communicator but if I'm going to spend the day with you, I need some hints of friendly engagement. But alas, I've worked with senior software engineers enough to know heart and head are not always aligned 🙃 It also wasn't driving me enough to next steps. For example, after I finished some nits in a technial spec, it never said: "ready to build"? These sorts of harness / prompt optimizations will make a huge difference when people decide what to reach for when coding. # Again, who are we building for? So Codex is a nice, opinionated app, focused on getting all the code that Codex models are capable of spinning out actually in production without too much babysitting, focused on the core development workflow of git branches and PRs. But as someone who seems to be the only person in the world who uses github desktop, I can't name another developer who has 15 worktrees going locally and also doesn't have a) claude code managing git via CLI or b) has memorized every git CLI command down cold. Codex Github GUI - better than GH Desktop App? Hi, it's me, the only user of the Github Desktop App Codex is a lovely if you're new to development and trying to learn the basics of code, git, skills, and automations. It might be a nice interface for PMs or designers pushing code. But OpenAI will have to "friendly" up the harness to match that persona, who has been trained to love that little lollygagging mascot in their terminal. # Build fast, learn things All that being said, the @OpenAIDevs is the best at engaging developers on early releases. Every bug, nit, weird response gets ack'd by the team, who was shipping iterative releases all through the week I tested. Sorry I love to test software I suspect we'll see Codex evolve over time, and I'm keeping an eye on harness updates. In the meantime, the Codex models will be staying in my collection of faves, and the Codex app can continue to try to lure me from the dock with every update.

Me last night: "Polly the @openclaw is worth $200k a year, easy" Me today looking at Opus spend: "Uhh 😳"

taste is simply being able to eyeball the correct corner radius to the 0.5px

I have found the perfect use case for @claudeai cowork: filling out a word doc that has been falsely marketed as a “form”

tfw you casually vibe code 85% of your friend’s VC funded startup over an afternoon

Them: vague posts about kicking of GPT-5.3 Codex and walking away for hours Me: screen sharing while typing “nooo why in the world would you do that?” as I yeet 93,000 lines of code over 5 days and tell Opus 4.6 I love her. Watch now: https://youtu.be/01zAtSYNlvY

it's wild how fast i went from search the docs to find the skills.sh skill skills.sh The Agent Skills Directory From skills.sh

I just built an infinite generative sci-fi story with 42 characters, and I have no idea how it ends. - MVP via @v0 - optimized by claude code - powered by @vercel AI gateway + workflows Honestly, I don't know what's actually happening in Sector 7: https://maplewood.zone

Question with AI is always: can you build with AI AND go fast AND keep high quality AND innovate? Luckily the @NotionHQ team shows us - absolutely yes. I had @brian_lovin, a product designer at Notion, show me his Prototype Playground - a deployed app where anyone on the team can create a namespace + build ideas for the Notion app that looks like it belongs right in the app. In this ep you'll see: - how to combine @usemonologue and @claudeai code to quickly build high quality prototypes - his flow for @figma to code, fast - and how he solved "icon hallucination" with a custom skill I love Brian's idea that ideas need to “encounter reality as early as possible," and this ep is his tactical playbook to achieve just that. A big big thanks to this week's sponsors: 💼 @WorkOS - make your app enterprise-ready today 🔀 @orkesio - the enterprise platform for reliable applications and agentic workflows Watch now: https://youtube.com/watch?v=s4HGbIhUgVo

lord give me the confidence of a LLM that thinks they can replicate a company logo in SVG

NEW: @chatprd will now index all your company features (live + planned) plus your competitors so you can vibe code your way into a better us vs. them feature matrix. May the SaaS collapse begin!

How I AI: Vibe coded myself a little evals / error analysis platform for @chatprd (co-contributor GPT-5.3 Codex)

"This is actually insane." - first comment on this week's guest + episode. @seejayhess is one of my favorite AI-powered engineers not only because he builds awesome DIY dev tools for himself, but because he shares those dev tools with the world. On this week's ep of How I AI, CJ shows us Flowy: part flowchart, part vibe-coded Figma replacement, which he uses to tell Claude exactly what he wants to build. You'll learn how - ascii and mermaid diagrams are not enough - you can "talk" back and forth with cc via diagram - his dueling model playbook for quality eng We also get a SURPRISE GUEST in the middle of the episode... you'll never guess who 🦞 As always ty ty ty to our amazing sponsors! 🧱 @orkesio - the enterprise platform for reliable applications and agentic workflows 🐶 @Atlassian Rovo- AI that knows your business Watch now on yt: https://youtube.com/watch?v=LC1mKvLWZ2E

mooommm the model providers are fighting over who is more evil again