Help Your Agents Help Themselves

AI agents are becoming increasingly capable of writing code, but building a good web experience takes more than that. They need to understand modern web practices, access the right tools to inspect browser behaviour, and know how to debug and verify what they build.

In this talk, Swetha, Senior Web Ecosystem Consultant at Google, and I will explore how to equip coding agents with up-to-date web knowledge and browser debugging capabilities. Through practical examples and live demos, we will see how agents can make better implementation decisions, investigate performance issues, and verify their own changes.

The goal is to move beyond agents that simply generate code, towards agents that can build, debug, and validate better web experiences.