Overview

Autonomy tests the journeys that matter before they reach your customers. You describe a journey as a few outcomes in plain words. An agent reaches each outcome on the live product, the way a careful tester would: it drives a browser or a device, calls your API, reads the test inbox, and waits for the SMS code. Every action it takes is shown under the step, with the evidence that proves the result.

When a signup, checkout, billing or onboarding journey breaks, you see which outcome wasn't reached, what the agent did, and what the screen showed. You don't have to rebuild the session by hand.

What Autonomy is for

Use Autonomy for product journeys where a unit test or a screenshot diff isn't enough:

• Signup, sign-in, two-factor, magic-link and password-reset journeys.
• Checkout, subscription and receipt journeys.
• Pull request checks against preview deployments.
• Mobile builds from CI or an upload.
• API journeys that call your endpoints directly.
• Release evidence you can share with product, support, QA and engineering.

How it works

1. Create a project for your product and add an environment: the URL, app build or API it runs against.
2. Write a test case as a few outcome steps, describe it in plain language, or record it once.
3. Start a run. The agent reaches each outcome and shows its Step Actions as it works. A Final check at the end of each step confirms the result.
4. Review the result: Passed, Failed or Couldn't start. Problems seen along the way are listed as issues, apart from the result.
5. Send results where your team works: GitHub checks, Slack, Linear, Jira, webhooks or your CI.

Where to start

Read the Quickstart to get to your first reviewed run, then Concepts for the terms used everywhere. When the first run is reliable, add pull request testing, schedules and integrations. Use AI assistants to create test cases, start runs and read results from your editor.