The bitter lesson of browser agents
The development of browser-based AI agents has evolved from rigid, rule-based action sets to flexible code-execution models. Developers found that allowing models to write their own code for browser interaction significantly improved efficiency and reliability.
Why it matters
This shift represents a fundamental change in how AI agents interact with the web, moving toward more autonomous and capable systems that can handle complex, real-world browser tasks.
As models get better, the browser harness has to change. A lot.
When we launched Browser Use in November 2024, GPT-4o wasn't trained for computer use. Built on next-token prediction, it didn't reliably understand what it could do on a page. We had to spell it out.
As models got better at coding, we moved from predefined state and actions to code: first the model wrote its actions, then its observations. Now it does both through raw CDP, inside an existing agent harness.
We defined the state: here's what's on the page, here's what's clickable. Then the action space: click, type, scroll. Every interaction had to fit those rules.
The model picked an action, we ran it and sent back the next state. But browser automation is a game of edge cases. Every exception needed another fix.
Get smarter about the news
Sign up free for a feed built around what you actually care about, Dive Deeper research on any story, and the full text of every article.
Create free accountAlready have an account? Sign in