From Telemetry to Tests: Building Agentic QA Loops with MCP and Playwright

The way we build E2E tests is changing. We are moving from static test suites and copy-pasted AI snippets toward agentic workflows where autonomous agents can explore applications, generate tests, validate assumptions and repair failures directly inside the development environment. In this talk we will explore how Model Context Protocol (MCP), Playwright Test Agents and runtime telemetry can be combined to create a real agentic QA loop driven by production signals and user behavior. Starting from a simple application, we will build a practical workflow where telemetry and real user flows are transformed into dynamic test specifications consumed by autonomous Playwright agents.