Skip to main content

Browser Agents Overview

Browser agents are autonomous bots that control a web browser to complete tasks described in natural language. Given a goal, an agent observes the current page (screenshot + HTML), reasons about what to do next, and takes action — click, type, navigate, extract. No scripts required.

How They Work

Each step, the agent:

  1. Captures a screenshot and reads page structure
  2. Consults the LLM with the current state and goal
  3. Decides on an action (click, type, navigate, extract)
  4. Executes the action and observes the result
  5. Repeats until the goal is complete or a failure is detected

The system prompt you write defines the agent's behavior and constraints.

Use Cases

  • Automated form submission and account workflows
  • Price and content monitoring
  • Data extraction across multiple pages
  • Regression testing after UI changes

Browser Agents vs E2E Tests

E2E TestsBrowser Agents
ExecutionFollows a scriptDecides in real-time
SelectorsCSS / XPathUnderstands element purpose
UI changesBreaksAdapts
Best forStable critical pathsComplex flows, dynamic UIs

Use E2E tests for CI gates and compliance paths. Use browser agents for exploratory testing, complex conditional flows, and workflows that change frequently.

Access

Browser Agents is currently in private beta. Access is granted on request — contact the team to enable it for your account.

Agent Lifecycle

PhaseStatus
Create, configure, run, view trainingAvailable
Labeling UI, evaluations, deployment versioning, production monitoringIn development

Agents start in draft status. Full deployment pipeline (versioning, channels, production monitoring) is coming.

Glossary

TermMeaning
System PromptNatural language instructions defining agent behavior
RunSingle execution of the agent
StepOne action within a run (click, type, navigate, extract)
Outcomesuccess / failure / error
Golden DatasetCurated training examples
LabelingReviewing run steps and marking correct/incorrect actions
Training RunA job that improves the agent from labeled examples
VersionSnapshot of the agent at a point in time
Channeldraft / candidate / deployed

Next: Creating an Agent