Plain-English tests
Write the workflow as readable lines of intent, with saved values and reusable actions when you need them.
TestPilot turns plain-English steps into real browser actions, visible checks, and review-ready evidence—locally or across CI.
Designed for the miles ahead.
The person who understands the workflow should be able to read the test. TestPilot keeps the scenario human, then uses computer use to work through the visible browser.
Create a uniquely named record, try to create it again, verify the duplicate warning appears, then clean up.
Start with the behavior that matters. TestPilot handles the browser journey and leaves a trail your team can review.
Describe the workflow as a sequence of outcomes. Reuse guided actions for the parts your team performs every day.
new test named "Order total stays consistent" open the product catalog add a random available item to the cart save its displayed price as "item price" open the cart verify the total equals "item price"
Write the workflow as readable lines of intent, with saved values and reusable actions when you need them.
AI reads the visible interface, then navigates, clicks, types, and verifies the same experience your users have.
Follow every running, passed, or failed step and keep the screenshot that explains what happened.
Run one browser per container, balance the suite across shards, and merge the results into one review-ready report.
Keep the instructions, the browser, and the evidence in one view— so the team can follow what the AI is doing as it happens.

Every shard contributes to one clean result: a browsable HTML run, machine-readable output, and screenshots that stay attached to the exact step that produced them.
Keep browser runs inside the environments you approve, separate checks from mutations, and stop stuck steps before they hold up the rest of the suite.
Tell us about the workflow you want to make simpler, more visible, and easier to trust.
Start a conversation