Overview

Autonomy tests the journeys that matter before they reach production. It can drive a browser, inspect mobile artifacts, verify API side effects, compare a screen against its design, wait for email or SMS messages, and turn the result into evidence your team can review from a pull request, Slack, or a release workflow.

Instead of separating browser replays, network logs, inbox messages, design references, and mobile artifacts into different tools, Autonomy keeps them in one run. The goal is simple: when a checkout, signup, billing, onboarding, or mobile release check fails, reviewers should see what broke and what to do next without reconstructing the session by hand.

What Autonomy is for

Use Autonomy for product flows where a unit test or screenshot diff is not enough:

• Signup, login, OTP, and password reset flows.
• Checkout, subscription, entitlement, and receipt checks.
• Pull request smoke tests against preview deployments.
• Mobile build validation from uploaded artifacts or CI output.
• API and network assertions that confirm the UI caused the right backend state.
• Contract assurance against a published OpenAPI document.
• Visual comparison of a live screen against the Figma frame it was designed from.
• Release evidence that can be shared with product, support, QA, and engineering.

How a run works

1. Choose a target environment, such as a preview URL, staging app, or uploaded mobile build.
2. Select or author a test case that describes the user journey in product language.
3. Autonomy executes the run with browser, mobile, API, email, SMS, and visual checks as needed.
4. The evidence workspace shows the live session, timeline, replay, artifacts, generated checks, and failure summary.
5. Integrations publish the result back to pull requests, Slack, or your release process.

Authoring a test case

Three ways, all ending in the same step editor:

• Write the steps when you know the exact sequence.
• Describe the flow and let Autonomy draft intent-level steps from a prompt.
• Record the flow — perform the journey once in a hosted browser and Autonomy drafts the steps from what it observed.

See Authoring.

Where to start

Read Quickstart to get to your first reviewed run. Use Pull Request Testing and the GitHub App when you want evidence on every preview deployment — including runs you ask for in plain language from a PR comment. Use Integrations to connect Vercel, GitHub, Figma, and Slack. Use AI Assistants when a connected MCP client should inspect cases, runs, and traces from your editor.