Browser Agent
An agent that navigates the web for you — opening pages, clicking, filling forms, reading results.
Think of it like
An intern you send off to “go find this online and book it,” who comes back with the confirmation.
Example
“Find the cheapest direct flight next Friday and start the booking” — the agent opens a travel site, searches, compares, and fills the passenger form.
How it actually works
Browser agents work over the page — sometimes via the DOM, sometimes via screenshots — issuing navigate, click, and type actions in a loop. They stumble on the modern web’s defenses: logins, captchas, pop-ups, and layouts that change under them. And because they act on live sites, prompt-injection from a malicious page is a genuine risk, so what they’re allowed to do should be fenced.
For product teams
The dream of “just do this task on the web,” bounded by how hostile the real web is.
For engineers
An agent that controls a browser via DOM or vision, emitting navigation and interaction actions against live pages.
Related
- Computer Use — The broader case of driving a whole computer.
- Hallucination — Malicious page content can hijack the loop.
- Search Tool — A read-only cousin that just fetches and searches.
Read anything AI without the jargon
Look up any term in plain English, or save terms as you read with the free Chrome extension.
Open DecoderAdd to Chrome