Finding signal on Twitter is more difficult than it used to be. We curate the best tweets on topics like AI, startups, and product development every weekday so you can focus on what matters.

Daily signal archive

Choose a day

August 18, 2026

Pick a date. Remember what happened.

231 posts98 authors154.0K engagement in top results
aakashgupta
@aakashgupta30d

Elon built a power plant in Mississippi to run computers in Tennessee. They're a few hundred feet apart. The state line between them is doin...

sama
@sama30d

We have paused some frontier RL training to ensure that we can meet the appropriate alignment, security and monitoring standards for the new...

lennysan
@lennysan30d

šŸ’„ Announcing Lenny’s Jobs: The best place in the world to find, vet, and land your dream job I’ve spent thousands of hours on my newslette...

shl
@shl30d

Code is mostly solved, but design isn’t. Introducing Tastelint! An agent that runs on every PR and surfaces design feedback. Reply below ...

claudeai
@claudeai30d

Claude can now send emails in Gmail and manage files in Google Drive. Ask Claude to reply to a thread, and it drafts and sends the response...

OpenAI
@OpenAI30d

As models become more capable, the risks associated with developing and testing them internally also grow. We temporarily paused reinforcem...

MengTo
@MengTo31d

I open-sourced the Sylva three.js site and the skills behind it. I started with one reference and asked Opus 5 to recreate it in Three.js a...

claudeai
@claudeai30d

Claude Cowork is now available on mobile and web for all paid plans. https://twitter.com/claudeai/status/2074525815820169320

bot
@bot30d

We've shipped several quality-of-life improvements to Grok Bot. Mobile notifications are now grouped by Bot and use their specific icon.

cursor_ai
@cursor_ai30d

We're making Git hosting more reliable, performant, and scalable. This post traces 20 years of Git infrastructure and explains how that his...

zai_org
@zai_org30d

GLM-5.3 API is now live. - Built for coding, defensive cybersecurity, and long-horizon agentic tasks - Priced the same as GLM-5.2 - Availab...

trq212
@trq21231d

go into CC and type /design <something you want to design> do it rn https://twitter.com/nateparrott/status/2089470636796059754

benln
@benln30d

Collected Grok Bot pro tips shared by the @bot team: 1) Connect multiple accounts across plugins (i.e. personal and work Gmail accounts) 2...

AnthropicAI
@AnthropicAI30d

Many drugs work by binding to a specific target in the body and blocking or changing what it does. An important first step in the drug devel...

dhh
@dhh30d

Every insult and objection to Omarchy is an echo of what I heard twenty years ago with Ruby on Rails. Hundreds of billions of dollars in ent...

aakashgupta
@aakashgupta31d

Google just bought 100 million emails and 500 million Teams chats from a dead airline for $10 million. Spirit shut down on May 2 with $8.1 ...

trq212
@trq21230d

weird that there's a "make a lot of money" button and nobody's pressing it (take your SaaS, make it headless, let agents use it, charge per...

signulll
@signulll31d

most ppl simply do not realize that before fb the internet was almost entirely anonymous. there was very little personal branding, in fact i...

dhh
@dhh30d

I sat with Lex for another 6.5 hours on Sunday to record part 2 of this. The podcast incarnation of "there are decades where nothing happens...

rauchg
@rauchg30d

Your software factory should be a monorepo. All your company context (design, marketing, sales, engineering, support…) in one place for agen...

Teknium
@Teknium30d

FYI if you're just learning about Hermes Agent - Hermes Agent is a 100% free, open source, MIT licensed agent harness that affords you comp...

aakashgupta
@aakashgupta30d

The internet is roasting Netflix's $24 million Indian air war show for its CGI physics. The real 1999 operation it depicts was the highest a...

vercel
@vercel30d

Agents can now exploit vulnerable sandbox boundaries, so we are testing ours in the open. $1,000,000 hacker challenge for Vercel Sandbox: ...

gdb
@gdb30d

we temporarily slowed scaling of our frontier training, including our largest planned frontier RL, to strengthen security and monitoring. we...

alliekmiller
@alliekmiller30d

NOT enough attention is being given to: 1) Claude dynamic workflows. Claude is literally generating a harness on demand! Agents spinning up...

dhh
@dhh31d

When laptops soon begin shipping with 4GB of RAM, you'll be happy to know that Omarchy still runs great. We already have many users running ...

rauchg
@rauchg30d

We are putting $1M towards verifying the security of Vercel Sandbox, in the open. You're free to test any model in the world to try and find...

bcherny
@bcherny31d

Small quality of life improvements like this add up. More on the way https://twitter.com/claudedevs/status/2089509659090780193

dhh
@dhh30d

Super + ` in Omarchy 4.1 will give you the classic Quake console pull down for your default agent. @tobi absolutely cooked on this one 🤘

thdxr
@thdxr30d

is it time to switch off github actions? what are the best alternatives?

devindesktop
@devindesktop30d

GPT-5.6 Sol is now 70% off in Devin Desktop and Devin CLI, through October 3. On FrontierCode 1.1, Sol achieves top-tier results. With this...

emollick
@emollick31d

My father just got a new MacBook after years with no laptop, and he did not know how to set it up He just told me that he took my advice to...

dhh
@dhh31d

There won't be an accessory or gadget or service left without an Omarchy plugin in about five minutes. This is OS development with hyperdriv...

jack
@jack30d

buzz projects (aka a self-sovereign GitHub alt) https://engineering.block.xyz/blog/projects-in-buzz

iannuttall
@iannuttall30d

I make more money from SEO and display ads than I do from my apps. There's something calm and enjoyable about using AI to research and buil...

Hesamation
@Hesamation30d

Qwen3.8 27B literally broke the SIZE → INTELLIGENCE curve. Look how far left it is in the optimal quadrant. Crazy that a 27B model is sitti...

Hesamation
@Hesamation31d

Umar just made a super long 19-hour video explaining distributed training from first principles (while visiting concepts like RoPE, KV Cache...

dhh
@dhh31d

The road map for Omarchy 4.1 is already several man years long... I suspect we'll be done in a few weeks šŸ˜„

gregisenberg
@gregisenberg30d

The biggest lie 12 months ago was the narrative that GPT wrappers were useless/you were dumb building a business on top of the LLMs, and now...

tannerlinsley
@tannerlinsley30d

How @Lovable moved https://Lovable.dev (400 routes, 850K lines, 42M monthly visitors) from Next.js to TanStack Start 🤯 https://lovable.de...

rauchg
@rauchg30d

I’ve been using http://fx.sh as my daily driver and it’s a one-way street. It’s 10-20x smaller than the major coding CLIs. It starts up ins...

mattpocockuk
@mattpocockuk30d

We need a concept like "grep hygiene" I.e. when your coding agent searches for a concept, it should receive relevant results Not a huge sl...

tdinh_me
@tdinh_me31d

Just uploaded the prompt + source on GitHub You can ask your agent to generate any kind of structure now! Demo: https://trungdq88.github.i...

awilkinson
@awilkinson30d

I don’t know how Wispr Flow is a business in 2027. Literally weeks away from not needing to exist. I have never seen a clearer example o...

garrytan
@garrytan30d

GBrain actually works with any AI Harness (Grok Bot, Claude Code, Codex, Hermes Agent, OpenClaw, OpenCode, all work) and any Postgres with p...

businessbarista
@businessbarista31d

Enterprise vibecoding is a shit show. Most companies subject themselves to one of two failure modes: • Blindly enable vibe coding. It becom...

nextjs
@nextjs30d

Instant Navigations in Next.js 16.3 lets you achieve snappy, app-like UX without giving up the benefits of Server Components. See how Cache...

AndrewCurran_
@AndrewCurran_30d

This is a new quote from Sam Altman to Alex Heath saying that the reason OpenAI is slowing training is because its unreleased models are sho...

alexeheath
@alexeheath30d

OpenAI is slowing down its AI training efforts because its unreleased models are showing ā€œvarious degrees of misalignment,ā€ Sam Altman tells...

danshipper
@danshipper30d

BREAKING: When today’s jobs are automated by AI, what will great human work look like? This is the most important question of our time. In...

huggingface
@huggingface30d

We've just surpassed 3 million models on the Hub šŸ¤— the community is accelerating towards an open, distributed future where open AI is ever...

aakashgupta
@aakashgupta30d

Atlas Cloud runs 400+ AI models behind one key, and their new launch is aimed at people who'd rather skip the wiring. Every few weeks a new...

theo
@theo30d

I love that all the replies are saying ā€œwtf are you sure you tried the same app?ā€ T3 Code has had its rough parts. We put it out as an earl...

AndrewCurran_
@AndrewCurran_30d

I've said before that Fable 5.1 has been ready to ship for a long time, for weeks. It may be live for some of us right now, a small stealth ...

creatine_cycle
@creatine_cycle30d

ā€œwow cursor origin is going to cook githubā€ ā€œhave you tried it?ā€ ā€œno. have you?ā€ ā€œno.ā€

petergyang
@petergyang30d

It's super annoying that Mac switches my default microphone and speaker source each time I connect/disconnect my Air Pods. I was going to v...

emollick
@emollick30d

The fact that LLMs have theory of mind at all is a huge deal (and was controversial when discovered), but you can still see the limitations ...

ClementDelangue
@ClementDelangue30d

Something super exciting happened quietly on HF over the past month: AI agents became AI builders, and they did it in the open! During our ...

dhh
@dhh31d

Grok is catching up so fast to the edge of the frontier. But what happened to Google? https://twitter.com/rails/status/2089445842994299311

aakashgupta
@aakashgupta31d

A Detroit suburb audited its own license plate cameras and found police in Houston and Dallas searching its residents' data over 10,000 time...

aakashgupta
@aakashgupta31d

For two decades, a YouTube view meant a human watched for roughly 30 seconds. Starting Sunday, it means the video loaded. The last time the...

MTSlive
@MTSlive30d

SITUATION UPDATE: OpenAI is expanding chain-of-thought monitoring for Sol-level models and above to scan a model’s internal activity at ever...

mattpocockuk
@mattpocockuk31d

It me We'll look back on "one dev, many terminals" as an awkward interregnum https://twitter.com/jaredpalmer/status/2082845336041587052

dhh
@dhh30d

"During this intro itself, I already installed it!" šŸ˜‚ Great tour of Quattro from @LinuxBTW. https://www.youtube.com/watch?v=qBKMe8AatY0

BrettFromDJ
@BrettFromDJ30d

made a thing for your next brand project. oddgrid — generative pixel grid art. pick a look, set your palette, add motion. export as PNG, v...

dhh
@dhh31d

There are now over 400 plugins available for Omarchy Quattro! We just launched this on Friday. Absolutely bonkers.

agazdecki
@agazdecki30d

Building a startup takes a little bit of crazy and an unreasonable amount of optimism.

AndrewCurran_
@AndrewCurran_30d

OpenAI says it paused reinforcement learning on its latest models "intended for deployment" for the last two weeks. This excludes internal m...

amasad
@amasad31d

This team doesn’t have ā€œAIā€ anywhere in their pitch but has AI growth rates and would have 10x the headcount if they weren’t so AI-pilled. h...

theo
@theo31d

It's legit trippy to use T3 Code on a bad network. You forget the connection is terrible until you alt-tab back to your browser and everythi...

aakashgupta
@aakashgupta30d

Jennifer Aniston earns more from Friends reruns every six months than Sadie Sink earned from ten years of Stranger Things. Sink's final sea...

AndrewCurran_
@AndrewCurran_30d

Detectors in ruins. This was always inevitable, one way or another. The moment models write at or above skilled human level they will be in...

thdxr
@thdxr30d

long running agents that can work for hours and hours to deliver unfathomable amounts of value are here https://twitter.com/kmdrfx/status/20...

emollick
@emollick30d

The problem is that the output from the AI detector fooling startup is bad writing. https://x.com/emollick/status/2089859591597142031/photo...

vasuman
@vasuman30d

If you’re a one person business, you should be setting up agents before hiring. You know the process the best. https://twitter.com/exm7777...

googlegemma
@googlegemma30d

What if your notebook could write back āœļø Check out this interactive journal running locally on an iPad. Powered by Gemma and Apple MLX, it...

awilkinson
@awilkinson30d

Having major issues with Grok @bot. I have about 15 bots and they now frequently say they’ll do something then flake or disappear. Can a...

AndrewCurran_
@AndrewCurran_30d

The pause in training does not appear to apply to Astra at all, if I'm interpreting this correctly: https://twitter.com/sama/status/20898054...

aakashgupta
@aakashgupta31d

The overnight stock market has already faced one real crash. It handled it by canceling $437 million of trades and telling 90,000 investors ...

linear
@linear30d

Rethinking delta sync at scale with @turbopuffer. Engineer @traverspw on how we made Linear’s sync latency predictable as workspaces grow: ...

petergyang
@petergyang30d

I hate digital clutter so I set up a Marie Kondo Grok @bot to clean up emails, files, and paid subs with my approval. Here's the prompt to ...

mattpocockuk
@mattpocockuk30d

Going to try bitter lesson-ing myself this week Instead of hand-rolling a deterministic loop to tackle tickets from /to-tickets I'll just ...

emollick
@emollick30d

If alignment issues are becoming big enough that OpenAI is willing to commit 20% of research inference compute to chain-of-thought monitorin...

eastdakota
@eastdakota30d

Thank you for coming to my TED talk. https://www.ted.com/talks/matthew_prince_how_ai_is_breaking_the_internet_and_what_to_do_about_it

mattshumer_
@mattshumer_30d

Is Claude down?

thdxr
@thdxr30d

i've never done a sponsored post but i'm finally willing to in exchange for cheap GPUs

kiwicopple
@kiwicopple31d

supabase docs at a glance

AndrewCurran_
@AndrewCurran_30d

French Public Accounts Minister David Amiel said in a press conference today that the French government intends to hire sovereign AI compani...

mattshumer_
@mattshumer_30d

Etched has shipped!! Insanely proud (small) seed investor. >$1T valuation incoming. https://twitter.com/etched/status/2089729087732605282

stripe
@stripe30d

Kraken is launching Krak Cards in the US, powered by Stripe Issuing. Spend crypto anywhere Visa is accepted. https://twitter.com/Krak/status...

steipete
@steipete30d

pssst, you wake the cli people that will give you $reasons why this can’t work. I was one of them before I saw the light. https://twitter.c...

EXM7777
@EXM777730d

Andrej Karpathy, OpenAI Co-founder, introduced the "LLM Wiki" - a knowledge graph inside Obsidian i built my own version for GTM and here's...

clairevo
@clairevo30d

I don’t care about intelligence. I don’t care about UX. I care that my agents aren’t annoying af.

NousResearch
@NousResearch30d

Hermes Desktop now has a beautiful, adjustable matte transparency mode

businessbarista
@businessbarista30d

Our engineers are always living 3-6 months into the future. They’re constantly experimenting with new tools and I've never seen them this e...

aakashgupta
@aakashgupta31d

OpenAI loses $1.22 for every dollar it earns and just filed for a $1 trillion IPO. Sam's "perfect context of your whole life" line is how th...

OpenRouter
@OpenRouter31d

Model pricing is getting more complex over time. Here's how DeepSeek peak-hour pricing works, in your local timezone: https://openrouter.ai...

agazdecki
@agazdecki30d

Your startup can survive more mistakes than you think so keep going.

aakashgupta
@aakashgupta31d

A $20 delivered burrito is real, and the burrito was never the expensive part. Start with a $10 menu price. Restaurants pad their delivery ...

emollick
@emollick30d

Some early evidence that AI may actually be accelerating discoveries in the areas where you would expect to see AI acceleration happening fi...