Richard Lee

@samtayyari @madsf88 Wow those are nice hangers

@madsf88 I filmed this at a wooden hanger factory in China. It was an unforgettable experience.

why is every man's dream to build a factory

@madsf88 Lego

@saranormous This is a great way to build conviction

is talking to a model fine-tuned on your own thoughts (say, for investment judgement) a sign of madness? given this kind of iteration can help models, maybe it helps humans "silicon ducking" instead of rubber ducking

When a model's internal thinking says "I should be honest...", I know I'm in trouble

Compaction is not as scary as it used to be

@nathancgy4 Khan Academy isn't dead!

Read through @nathancgy4's paper on Attention Residuals... Reviewing my lin alg this morning...

@hthieblot Darn I have 24

The next billion-dollar founder has 22 followers on X rn. I will find you & fund you!

As we get deeper into model governance, organizational hierarchies and roles make sense for controlling the responsibilities and access agents have. We can't keep granting permissions one by one

@elonmusk Can you make Imagine generate across a disttibution of human intent rather than activations since I find the limitation with inage gen is in my specificity

Minute-long story made w Grok Imagine

I already love using AI studio for quickly prototyping designs but this is just crazy

Learn more: cursor.com/blog/composer-2

Tomorrow we will unveil the all new vibe coding experience in @GoogleAIStudio, the team has spent 4 months rebuilding it all from scratch and smoothing out rough edges to help everyone bring their ideas to life. This is a big step forward, but just the start : )

Had a call with Cisco leadership today to check in on their Codex deployment. They’re leveraging AI coding and codex more aggressively than most startups and frontier tech companies I talk to. Skills for everything, agent swarms to execute bigger projects, automated code review, deep system integrations, and more 👀 They’ve quickly hit new bottlenecks that we’re jumping in to help them solve with more codex. If your public company exec team isn’t pushing the org to cut project timelines by 95%, maximally leverage coding agents, and reimagine the SDLC, you’re ngmi. Shoutout to Jeetu Patel - bullish on Cisco.

Large scale agent simulation is a super cool space because activating LLMs in different ways produces drastically different results. I don’t believe you can replicate that with single agents

The Oasis Paper uses the public Twitter algo from 2015 so it no longer matches the current LLM driven algo

I dug into the source code for the simulation engine github.com/666ghj/MiroFish today. Ir’s largely a UI around getzep.com context graph and github.com/camel-ai/oasis for simulating large numbers of agents

I think suno.com is massively underrated

It still drastically underperforms Opus 4.6, but for the cost it’s unbeatable

GLM-5-Turbo is a testament to how fast and broadly China has adopted Openclaw. It’s fine tuned specifically for Openclaw with tool calling, instruction decomposition, and scheduled tasks