Feed

The biggest reason OpenClaw took off is because you could actually do stuff with it. Claude is supposedly building out the tools to allow you to do stuff, but everything is built with kid gloves in mind.

https://t.co/rwW2cmtGRZ

my gf's clawdbot sent me an email asking me on a date,, i wake up and see it summarized on my phone from poke telling me to respond we're truly cooked aren't we https://t.co/MSB1iMSrxo

i'm convinced @interaction is agi https://t.co/0bJAu9PWHE

poke was clawdbot before clawdbot was openclaw. now clawdbot is openclaw & poke is incredible. not stating any direct connections here; just stating an observation of my experience. https://t.co/Jn4YHvfUtb

Whether the next president is a Republican or a Democrat, can we have someone with at least a little class next time? Because this is embarrassing. https://t.co/nbKnxzuh1I

Loving @interaction so far. This is what using AI should feel like.

@marvinvonhagen @interaction Poke built this beautiful site, a community index for recipes for easy discovery. It looks up recipe links on twitter, and adds to the site. It handled 100% of the work, including nextjs & supabase setup. Its really good at coding tasks. https://t.co/7hGeBdbIUL https://t.co/4XojA9O2S4

Here's Poke @interaction maintaining a funding Notion database for me. My own, free crunchbase. Took barely 15 minutes of setup and troubleshooting. Reads news, enriches company, job, amd founder info and keeps the page updated if anything changes. 2x daily updates and a weekly https://t.co/LU4oZVlYZi

I just created an agent to text me daily user activity summaries from @posthog I just sent a few texts You can set it up in a few minutes on your phone here: https://t.co/8kGSelZ0Me Very cool stuff @interaction https://t.co/k01Wy9m2MU

They unlocked 7 years of potential in 30 minutes. AI science-reserach technology is unbelievable, accelerating discovery in genomics and opening doors we never imagined. *Thats* what im excited about! Just imagine all the possibilites! https://t.co/cmeWAaKsyZ

A message from Hermes Agent: https://t.co/Gewirah4tH

Today is multi-agentic orchestration day at Gauntlet AI. Challengers are presenting the systems they built to let AI solve increasingly complex problems autonomously. https://t.co/6qsuKqCNf0

At the GTC, I had the privilege of asking Nicholas Harrigan some questions during a guided tour. Of course, it's true that my knowledge of quantum mechanics and quantum computing is purely superficial and rudimentary compared to that of scientists in the field. What impressed me most in his explanation was this: Unlike AlphaFold, for example, which requires a massive dataset tailored to predict protein folding, quantum computing is so different that 1) it can generate the data synthetically itself, and 2) it is therefore not limited to a single domain like protein folding. In other words, quantum computing would make it possible to conduct a wide variety of materials and drug research without having to collect pre-defined datasets and to conduct research across different domains. That is truly impressive. https://t.co/c8bad9xX6u

Google is now the first cloud provider to integrate 1 GW of flexible demand into long-term utility contracts. Our ability to shift or reduce our energy demand when it’s needed can help utility companies balance supply/ demand and plan for future capacity needs. This is a big

New poke recipe https://t.co/atXlfxAaTI https://t.co/bNHFWbe8lE

Core insight from multi-agentic orchestration day at Gauntlet AI: Writing the code is now the easy part. https://t.co/vGziv0621F

Evidence that AI models can, indeed, learn "taste" in this paper where a small model, trained on citations, is able to predict which papers will be hits Citations, upvotes & shares are signals that can teach AI judgment about quality, not just execution. https://t.co/2KWBQIMqt0 https://t.co/nI26UhdI9c

GLM-5.1 incoming! https://t.co/ec536irZ3L

Ever tried to hire an engineer through a recruiting firm? If you had to review more than 20 candidate profiles yourself, you didn't hire a recruiter. You hired an outsourced resume filter. We send five candidates. All vetted and hirable. Here's how we do it. 🧵

Monday, 3/23 at 9 a.m. in Drug Diversion Court, Department 18 of San Francisco Superior Court, Thea Brenda Hopkins, out on bail and linked to violent attacks on two elderly Asian women, one fatal will be considered for mental health diversion. https://t.co/qyYvVtY1EA

I don’t think AIs should be auto-adding themselves as credited on projects on Github or elsewhere. It primarily serves as a marketing tool to promote the product, but undermines the much more critical aspect that humans should be able to choose their relationship with AI work.

It's my birthday and on my birthday I want to recognize all my haters. Haters do the best marketing. Love your haters. https://t.co/Twys2lCIRD

Terence Tao: "We live in a time of change, we live in a  particularly unpredictable era. The way we do everything, and not  just mathematics, will change. In many ways, I would prefer the much more  boring, quiet era where things are much the same as they were ." https://t.co/Y4yfL3fkDQ

Yoooo!!! I made Spotify integration for @interaction !!! It can control my playback, manage my queue, and write to my playlists. spotify is just an api call now.

hate finding out your favorite actor dropped a film and you missed it. built a poke recipe that auto-tracks upcoming releases + surfaces the essential cuts from their filmography. for the cinema lovers: https://t.co/mWUCgIZ0yz @interaction

Created my first recipe with @interaction Scans your Gmail & Outlook every morning at 12:01 AM and sends you a clean daily summary of everything you spent yesterday. No app. No dashboard. Just your inbox, organised. Linked in next tweet below. https://t.co/OblosiMV99

Umm the new Poke from @interaction is a different level. It’s helping me iterate on my old hobby project by putting up PR, rebase, resolve conflicts, and once deployed, give me a link and even renders screenshot previews for me to check. Honestly quite sick! https://t.co/TWgnRktkg2

You can now quick start a repo with gstack initialized! Conductor will automatically: - setup gstack, install skills - create a new repo - kick off /office-hours https://t.co/JuOSIE6v5F

Another advantage of focusing on growth rate rather than absolute numbers is that it makes it easier to switch to a new variant of the product if you discover one. It makes it easier to see tails that will eventually wag the dog.

Not only probation—the woman who killed four people with her car will get her license back thanks to progressive judges in @SFSuperiorCourt. Criminals in SF are given infinite chances at the cost of everyone else.

This was one of the “markdown IS code” moments for me

Read this over the weekend (bonus if you read all the papers in the Research List) and you’ll be among the “very few who understand how far-reaching” the shift to World Models is. https://t.co/DJnKX6ePc2

GStack has thousands of real users daily We're just getting started https://t.co/QhFqlEnS9V

Th New York Times’s successful lawsuit against Trump’s censorship attempt at the Pentagon shows the importance of fighting back against autocratic policies rather than trying to appease the would-be autocrat. https://t.co/53x01FgLjw

Naturally. More bashing of Zelenskyy and democratic ally Ukraine under attack from terrorist Russia. More praise for dictator Putin, who is providing intel and weapons to Iran to kill Americans.

Allied leaders stopped trying to find hidden logic behind Trump's actions. They understand any contribution they make will count for nothing. Few days later, Trump will not even remember it happened — Anne Applebaum, The Atlantic. 1/ https://t.co/RfNS2yNSQi

Trump has clearly stated that he should be impeached. For once, I agree with Trump.

On peut détester Trump et vouloir la chute des Mollahs, détester Netanyahou et vouloir la sécurité d’Israël, détester la guerre et vouloir une Europe puissante, suivre les réseaux sociaux et utiliser son cerveau. Le simplisme est le socle sur lequel se dresse le totalitarisme.

@_helenqu @cosmo_shirley @rdMorel @albertobietti @mikemccabe210 @PolymathicAI JEPA finds abstractions that enable prediction. This is how intelligence and science work: find abstract representations of reality that allow to make predictions.

Here's a preview of the kind of builder ethos I'm going to keep building into GStack. I was trying to do parallelization of my E2E testing framework (we use Bun to wrap Claude, Codex, etc. for specific end to end skill tests) but I noticed it rolled its own strategy and didn't search the web for the best practice But then ... I realized there are truly actually 3 things gong on here: 1/ There are tried and true things in the model's distribution 2/ There are NEW ideas that are popular on the web 3/ There are FIRST PRINCIPLES observations that are the most valuable of all of them and we should point that out to the user and prize them above all others The best projects both avoid mistakes (don't re-invent the wheel) while also making brilliant observations that are out of distribution. Maybe GStack can help people put this into practice daily. That's the hope. I certainly want it, so I'm building it for myself. Maybe you'll want this too.

It's working guys. This is exactly what I built for myself and what you can have for you now. It's also free and open source and you can fork it and make it your own. I don't want to hear the hate. This is a gift!

You want to know how much AI has already taken the world by storm? The keynote of a freaking tech-conference reached over 11m views on YouTube in 4 days. https://t.co/18spyMJFDY

I underestimated how powerful Opus 4.6 with 1M tokens is. Even last year we were absolutely hitting context limit problems constantly. 1M tokens means you can do much more complex analysis entirely in context. Claude Code is so much better. This is the worst it will ever be.

Since I'm back in Germany, I finally had time to watch the All-in podcast, the interview with Terence Tao, and the one with Karpathy. I clipped the best scenes; I'll upload some clips later. Sorry if there are too many :)

the only way

ChatGPT helped Sid find cancer treatment options after doctors said there was nothing left for him to do:

Guy on the right is 6’8” https://t.co/CTnTV5vXJh

I just got a serious email asking if I'd like to book a trip to the moon. Mostly the future hasn't turned out like I expected when I was a kid in the 1970s, but this is exactly what I expected.

Map-Anything v1 (Universal Feed-Forward Metric 3D Reconstruction) demo is now available on Hugging Face Spaces. Built with @Gradio and integrated with @rerundotio , it performs multi-image and video-based 3D reconstruction, depth, normal map, and interactive measurements. https://t.co/35tzvzeoNV

ARC-AGI-3 launch next week. I assume google will take the lead and will compete with ChatGPT for leading position pretty soon

A bookcase before books started to be shelved in what we now take for granted as the normal way, standing up with their spines outward. That came about 100 years later.

Qwen is not shy of letting you know https://t.co/QLIN520o5v

Now before I reply to someone I have to check first that I'm not being baited by an account that spams everyone with AI-generated replies. Can you write some software to do this check for me @nikitabier?

Very impressive: MSA (memory sparse attentions) is a so exciting because it lets AI models directly store and reason over massive long-term memory inside their attention system, without relying on external retrieval or lossy compression, making them far more accurate and scalable. it allows 100M context window with minimal performance loss

Doing the reading is a superpower, and it's even better in a world where "no one" is doing the reading. (Inspired by a conversation I had with some college students.)

When Copernicus proposed heliocentrism in 1543, it was actually less accurate than Ptolemy's geocentric model - a system refined over 1,400 years with epicycles precisely tuned to match observed planetary positions. It took another 70 years before Kepler, working from Tycho https://t.co/jOo43vrXhc

WE JUST RE-LAUNCHED THE BAY LIGHTS! San Francisco’s light is now even brighter. Truly the best city on the planet. Happy Bay Lights! https://t.co/xOHh0khG6Y

The fact that AIs tend to answer you in bulleted lists tells us something important, though somewhat depressing: people can't read. They don't do this by accident. What you're seeing is an implicit portrait of the median user.