Most automation teams have spent years building solid, well-structured test suites, hundreds of tests, established conventions, and genuine engineering investment. The arrival of agentic AI has left many teams asking the same question: does adopting it mean starting over?

This session argues the answer is no, and demonstrates why with a live, unscripted example. By connecting a GitHub or Azure DevOps repository directly to a multi-agent AI platform, it's possible to ingest an existing codebase and automatically map its Page Object Model structure, naming conventions, tags and coverage patterns, without any migration or rewrite.

The result is a static test suite transformed into a conversational interface that the whole team, technical and non-technical, can query in plain English. From there, agents can execute tests via natural-language commands, self-heal failing scripts, surface coverage gaps, and generate new Playwright scripts that match the repository's existing style, all triggered from a single user story.

This talk walks through that transformation live, using a real repository rather than a sanitised demo, to show what a genuinely conversational, self-maintaining test suite looks like in practice, and how existing automation investment becomes the foundation for it, not an obstacle to it.

Attendees will learn:

- How a single repository URL can unlock full agentic test orchestration without migration
- What working with a conversational test suite feels like day to day
- The multi-agent architecture behind contextual intelligence and autonomous remediation
- Why an existing test suite is a team's richest AI asset, and practical steps to uplift it