With T2T Autopilot you write the test instruction in plain text exactly as you would describe it to a colleague. No syntax. No selectors. No step-by-step breakdown required. You paste it in, and the LLM takes it from there.
It reads the screen, reasons about the flow, and executes the steps dynamically. If something on the page looks different than expected, it adapts. It does not fail on a missing class name.
A test case in Text2Test is not a single instruction type. It is a sequence. Each step can be a different prompt type, combined in any order to match exactly what you need to test.
Write the steps in plain text. Text2Test Native reads the screen semantically and executes. No selectors, no DOM knowledge required.
Paste an existing script or auto-generate one and run it inline within the test. For when you need precise programmatic control at a specific step.
Hand control to AI for complex flows that need reasoning. Autopilot navigates the product autonomously and hands back when done.
Take a screenshot diff or check element visibility at any point in the sequence. Catches layout regressions and UI changes.
Call an API and assert on the response. Combine with UI steps to test the full stack in a single test case.
Mix and build in any order. A test case can start with a T2T Native step, run an HTTP request to set state, verify visually, inject a script for a specific action, and hand off to Autopilot for the final flow. All in one sequence.
Request a demo and we will show you T2T Autopilot running a real complex flow from a single plain text instruction. No script. No selectors. Just reasoning.
Request a Demo