Pete Koomen (@koomen) 's Twitter Profile
Pete Koomen

@koomen

GP @ycombinator, cofounded @optimizely

ID: 14701449

calendar_today08-05-2008 15:21:56

1,1K Tweet

7,7K Followers

1,1K Following

Michelle Tandler (@michelletandler) 's Twitter Profile Photo

I've been in San Francisco for a month and can confidently report that the city is *back on track*. The streets are clean. The tents are gone. I've seen just one pile of glass. SF's collapse was the direct result of Progressive leadership & policy. Now, we recover. 🤍🌁

Mike Knoop (@mikeknoop) 's Twitter Profile Photo

We found this idea of user-owned system prompt is a requirement, not an option, for Zapier Agents. There is no "god prompt" that works reliably across all use cases. The only way to get Agents reliable enough to deploy is let end-users locally steer to their specific use case.

Pete Koomen (@koomen) 's Twitter Profile Photo

I'm excited to return to my alma mater, UIUC, on May 9th to talk to builders about building. Follow the link in this post to join the waitlist for my talk and/or 1on1 office hours! Thanks to buildillinois for hosting ❤️

Suhail (@suhail) 's Twitter Profile Photo

You know the app layer will win because instead of saying <model did it>, users will say <app did it> and what they will recommend is the app, not the model. The app will take extra special care to solve their problem.

kalomaze (@kalomaze) 's Twitter Profile Photo

isn't it kind of absurd that someone can just push a inconspicuous single commit to change the typical experience of a tool that ~122 million people use every day in a way that will have qualitative changes on the tool's behavior that are not robustly interpretable or measurable

Tom Blomfield (@t_blom) 's Twitter Profile Photo

When building AI tools for highly trained professionals like doctors and lawyers, it’s useful to give the user control of the system prompt. Letting users define how the agents work makes AI feel like an extension of themselves. They bring their style, their know-how and their

John Yeo (@johnyeo_) 's Twitter Profile Photo

We've just launched pricecn -- a free and open-source collection of beautiful shadcn pricing components After collison-installing autumn for many customers, turns out the most tedious part about billing is all the nitty gritty frontend work We plan to make this 10x simpler

We've just launched pricecn -- a free and open-source collection of beautiful shadcn pricing components

After collison-installing autumn for many customers, turns out the most tedious part about billing is all the nitty gritty frontend work

We plan to make this 10x simpler
Aaron Levie (@levie) 's Twitter Profile Photo

The reason why going AI-first as a company is so important now is best captured by Paul Graham's startup advice of "live in the future, then build what's missing." AI-first companies will see new problems to solve soonest, and that flywheel will only compound over time.

shadcn (@shadcn) 's Twitter Profile Photo

Introducing the registry mcp. One command to make any component registry mcp-compatible. Your Design System. Now with AI. Zero config. We've got a lot to cover. Let's get started. ⬇️

Introducing the registry mcp. One command to make any component registry mcp-compatible.

Your Design System. Now with AI. Zero config.

We've got a lot to cover. Let's get started. ⬇️
Pete Koomen (@koomen) 's Twitter Profile Photo

Congrats to Tom Invernizzi and Arthur on launching Combinely! These two are using AI to help accountants save time on 'quick' questions from clients that consume 30-40% of every working day. They're solving a problem they understand well thanks to Tom's accounting experience.

Mike Solana (@micsolana) 's Twitter Profile Photo

frustrating when people incoherently credit san francisco's slow turnaround to "AI" (???), rather than a well-funded years-long political strategy that sought and then delivered several recalls, a new DA, a moderate board, and a new mayor

Harj Taggar (@harjtaggar) 's Twitter Profile Photo

Optimizing AI agents requires writing down the valuable institutional knowledge in your company. Big co’s will struggle with this: middle managers worry about becoming replaceable and leaders worry about leaks. If they don’t do it, they’ll be crushed by competitors who do.