Live LabsBuildersBrowser agents2 hours

Build a browser agent that finishes the task

Browser agents demo well and break quietly. We build one that finishes a real multi-step task, with stop rules and a recording of each run.

You leave with

An agent that completes a multi-step web task, and a way to see where it fails.

How the time is spent

  1. I do30 minParam builds it live

    I run a live multi-step task on a real site and replay where it went wrong last time.

  2. We do48 minTogether, step by step

    We add a custom tool and a stop rule, then rerun the failing case.

  3. You do42 minOn your own code

    You automate a form or lookup on a site you use every week.

What you will be able to do

Pick the right control

Know when to script steps and when to let the model act.

Make it stop

Add limits so a lost agent does not loop.

Debug a run

Replay a failure step by step.

Starts from

Before you come

For
Engineers automating back-office and research work in the browser.
You need
Python or TypeScript.
Stack
Python, TypeScript, Browser Use, Stagehand
Track
Hands-on agentic AI for engineers who ship.

More builders labs