End-to-end tests are powerful, but they can be slow to write and easy to drift. We start with requirements, turn them into test code, and over time the original intent can get lost. What if we could keep that intent in plain language and give AI agents a repeatable way to turn it into reliable E2E coverage?

I’ve been using playbooks to give AI agents a repeatable way to work through E2E testing. The playbook tells the agent how to gather the right context, turn requirements into clear scenarios, use Playwright CLI to work through the flow, and validate the result. The goal is not to generate more tests. It is to give the agent a repeatable process for creating the right ones.

In this session, I will show how the playbook works in practice, where AI helps, and where people still need to stay involved. We will look at how requirements, designs, and existing code can feed into the process, and how the results are checked before they become part of the test suite. You will leave with a practical way to use AI for E2E testing without losing control of the behavior your tests are supposed to protect.