it's fascinating that we can't even decide how well LLMs work. Even for relatively well-defined problems.

I’ve seen 4 posts today across 2 social networks saying “what’s the cursor for X” where the answer is cursor

i’m just saying if i were an 18th-century weaver, and loom-powered factories were popping up i’d try to get really good at using looms

when figma mcp server?

me, while cursor explains everything it changed in my codebase

whoa did y’all know the Ghibli generator can also answer your questions too?

can’t believe that the AI intelligence debate finally went mainstream because people saw a ghibli of it

I’ll bet at least 80% of the people hyping MCP have never used an MCP server

Free startup idea of the week: Ahrefs for LLMs is anybody doing this well?

You ever think about work in the 70s and 80s and wonder how they got things done without computers? Our kids will wonder how we got things done without AI

Google is finally getting it.

yeah we haven’t cracked LLM memory yet

The day when no one can trust a video is threshold moment. People are sleeping on the significance of this.

Humbled and pleased to announce I made it to p.176 in the meeker deck

“What if OpenAI builds this” is scarier than “What if Google builds this” ever was

Every data scientist on X right now is so happy we’re talking about titanic survivor rates

Consistency is a weapon. In everything you do. It’s shocking how few people really get this.

Prediction: we’ll observe an upward bend in life expectancy in 2024 because of LLMs

is mcp server the new ICO?