A browser agent is an AI system that uses a web browser for you, opening pages, clicking, typing and reading, to complete a task you describe.
In plain words
A language model on its own can only write text. A browser agent connects it to a real browser: the model sees the page, through its structure, screenshots or both, and picks the next action, such as click this button or type in this box. It repeats until the task is done. Because a page can carry prompt injections and actions can have real effects, everything a page says must be treated as untrusted.
A quick example
Imagine asking an agent to find the cheapest train ticket for a date. It opens the booking site, fills in the stations, reads the results, and reports back. Tasks like this are hard: in the 2023 WebArena benchmark, the best GPT-4-based agent finished only about one task in seven.
Where you'll meet it
- Automating repetitive tasks on websites.
- Filling in forms and collecting information across pages.
Related
Sources
- DocsBrowser use tool, Anthropic Claude documentation · read 5 Nov 2026
- PaperWebArena: A Realistic Web Environment for Building Autonomous Agents, Zhou et al. (arXiv) · read 29 Sept 2026
- Docsbrowser-use/browser-use, Browser Use (GitHub) · read 15 Oct 2026
Checked by WhatIsAI against the sources above · 5 Nov 2026. Spotted an error? Report it.