We killed your startup, lol—what to build when Sam Altman keeps shipping better models

What do you build when the models just keep getting better? I wrote a guide for builders, startup founders, and frankly anyone who is worried these days about their job because they work at a startup!

A couple of days ago I wrote about strategic frameworks guiding AI players in 2025. That post was mostly focused on how to correctly read and build against core strategic dynamics shaping AI, but it didn’t really address the elephant in the room for indie builders and startup teams:

Yes, this is a real slide. Yes that’s Sam Altman in front of it. From way back in 2023. This has been happening for awhile. Read on to find out what we should be doing about it.

Sam’s own advice is just to build with evolving intelligence, which is pretty vague and also correct. My view is that a simple directional statement like that isn’t good enough to really put meat on the bone for anyone concerned about products—which is, like, most employees in tech at this point because we’re all affected when a model drops.

So I wanted to put something together that’s much more useful if you’re interested in how to avoid getting stomped on by OpenAI, or Google’s next Gemini drop, or DeepSeek dropping R2 (maybe this week?), or name-your-model.

I’m interested in how we can think about this deliberately, so we have a frame to fall back on when the next model releases tomorrow or next week and it’s really really good, again. And it changes the game, again.

Basically, we’re all playing a board game in tech where a few people are changing the game unpredictably. But this is an open-world game. We can find our own spaces. So how do you play a game in this world where the most important rules don’t get reset all the time?

That’s why I wrote this. You can call it a guide to product, but you can also read it as a guide to where to work—the best startups to work for are already doing this. Enjoy!