oh you’re still doing prompt engineering? everyone’s on context engineering now. just kidding, we’re all about agent design. we were using multi-agent swarms, but then the devin guys published that blog post saying not to, so we pivoted the whole stack to a single-agent

i think the new meta will be to take already existing software companies and to recreate them 100x faster 10x cheaper for the end user. software is easy to recreate once you have an already existing example of a working product.

Generalist robots don’t fail due to a lack of generality. They fail due to a lack of proficiency where it matters. Today, we introduce SOP (Scalable Online Post-training)—a framework that enables generalist models to evolve continuously from real-world experience across

there it is- "today we're introducing Personal Intelligence" now your emails, photos, youtube & search history, location, documents will all be used to train a personalized version of gemini to deliver you a tailored experience. this is all part of googles multi-pronged

GM 🌊 The tide is rising. We are thrilled to announce that the Abyss Beta will be rolling out in the coming weeks. We’ve been building in the dark, and it’s finally time to dive in. 🧵

We are live... beta.abyssprotocol.xyz

The BSA has been working behind the scenes to bring something massive to the community, and it’s finally here. We’re hosting our next conference “Stablecoins & Payments”, a deep dive into the tech actually moving the needle in 2026. Stablecoins are the bridge between the

Today is my last day at Anthropic. I resigned. Here is the letter I shared with my colleagues, explaining my decision.

Chrome 146 includes an early preview of WebMCP, accessible via a flag, that lets AI agents query and execute services without browsing the web app like a user. Services can be declared through an imperative navigator.modelContext API or declaratively through a form.

just read this AI article and something broke in my brain that i can’t unthink of crypto was never for us. we're just the beta testers who showed up early.. some thoughts: what does AI need to function as economic agents? > way to receive payment (they provide services,

Someone just built a fully functional AI assistant in 1 day.. and it runs on 10mb ram. It's called PicoClaw and it uses 99% less memory than OpenClaw. 100% open-source.

people giving an llm full unauthenticated access to your system are peak 2026 tech bros. the clanker can run an rm -rf just because an unread spam email told it to you guys are running a personal assistant that has 512+ critical vulnerabilities. 2026.1.15 patch added a backdoor

A new scam plague is unfolding, mainly targeting agents accessing your emails and docs. From malware → botware. Agents will be tricked, hijacked, and exploited at scale. This will be massive. Good luck.

"The most urgent film of our time." THE AI DOC: OR HOW I BECAME AN APOCALOPTIMIST is only in theaters March 27. Watch the trailer now.

We have raised a $110 billion round of funding from Amazon, NVIDIA, and SoftBank. We are grateful for the support from our partners, and have a lot of work to do to bring you the tools you deserve.

I spent time in Shenzhen last year and when I saw Merz come back from China saying Germans need to work more I immediately knew what broke his brain because I lived the exact same cognitive shock my first week in Huaqiangbei I burned through 4 prototype iterations of a motor

Our newest model, π0.7, has some interesting emergent capabilities: it can control a new robot to fold shirts for which we had no shirt folding data, figure out how to use an appliance with language-based coaching, and perform a wide range of dexterous tasks all in one model!

Arkworks 0.6.0 is out 🎉 The headline isn't BabyBear, KoalaBear, Mersenne31, Goldilocks. It's `SmallFp` — a macro for any prime up to 64 bits that picks the optimal layout + arithmetic at compile time. Drop-in. Non-breaking. Up to 30% speedup. READ: andrewzitek.xyz/small-fields

well done 👏🏼