Pinned Tweet
Five years at the spaceship. Wins, losses—growth. The greatest professional privilege of my career. Grateful for the people, the lessons…the mission.
Just getting started.
1
29
1,340
I despise stop and frisk Slop Policing in the same vein I despise dumping slop on your teammates without using basic hygiene and etiquette:
1. British spellings.
2. Things that are clearly misleading or incomplete.
3. Ideas without local context.
4. Erroneous info.
A new faux pas I’m seeing: dismissing a teammate’s well-structured, probably decent agentic output as “AI slop” because you don’t want to read it.
If a teammate (or agent) has skillfully prompted and carefully edited something to solve a problem or communicate with the business, it isn’t automatic slop just because AI generated it. There’s often good stuff in there... if it’s long or you're multi-tasking, feed it to your agent.
Don’t overuse the term "AI slop" for any AI output. Save it for the actual sloppity slop, ya know?
148
Put “have you no decency” on a hoodie bruh, let me cop one💀
This Post is from a suspended account. Learn more
167
During pre-release testing, Claude Fable 5.2 and Claude Mythos 5.2 worked on ambitious projects in many different fields. Here are a few.
From a one-paragraph prompt, Fable 5.2 designed three working electronic devices, including this home voice assistant hub.
The future of intelligence is small, it’s private, it’s sovereign.
Andrej Karpathy:
“A 1B model can be much more effective than most current LLMs.
90% of current models waste a ton of capacity and money on remembering stuff that doesn’t matter.”
here is his 5 core thesis about Small LLM’s:
1. the intelligence was never the knowledge. Strip the memorized internet, keep the reasoning. What’s left is the cognitive core, and it’s small.
2. the core might fit in ~1B params. Maybe less. “Plenty of room at the bottom.” We’ve been paying for a library when we needed a brain.
3. size stopped being the bottleneck. What makes an agent work is the tools it can call and how tight the feedback loop is, not parameter count.
4. intelligence lives in the system, not the weights. A raw LLM is a CPU with no OS. Small core + good harness > giant model alone.
5. you can own it. A model on your hardware is one no one can rate-limit, price-hike, or deprecate on you.
in 12 months the gap between teams shipping on-device and teams paying per-token frontier bills won’t be capability.
TwIL an 8.8B model at 90%+ of frontier isn’t a benchmark flex.
It’s the thesis Karpathy’s been repeating all year, made real.
study more on this below ↓
1
139
I’ve literally never called it anything but its government name, what’s on it’s birth certificate.
3
132
Jazbo retweeted
The API is super cool and I think you're gonna love it!
Big news for our developers! We are shipping a developer preview that dramatically expands the capabilities of all apps to integrate into Siri. So excited to see what you folks do with this.
developer.apple.com/news/?id…
Jazbo retweeted
There's a buzzy new paper about GPT-6 Astra being able to directly interpret model weights (prompts full of floating point numbers) - I tried it with Qwen 3.8 27B on my laptop and found that, after 45+ minutes, that model could achieve the same thing simonwillison.net/2026/Nov/3…
“No ML background required.”
Applications are now open for the Anthropic Residency, a paid, fixed-term research program for academics, technical founders, and staff-level engineers who want to make AI safe and beneficial for everyone. No ML background is required.
Apply by November 22: alignment.anthropic.com/2026…
1
1
174


























