AI Testing Inside the Developer Loop
Wei-Wei Wu, CEO of Momentic, on AI-native end-to-end testing
The winner in AI testing will be the product that lives inside the developer loop, not beside it. For end to end testing, the important work happens while code is still on localhost, inside VS Code or Cursor, and inside pull requests before merge. That is why GitHub bundling Copilot into VS Code and Azure DevOps is strategically strong, but it still leaves room for tools like Momentic that run locally, plug into GitHub, and turn test intent into a blocking CI check with less maintenance than Cypress or Playwright.
-
Momentic is built around a local runner and desktop app, with tests stored in GitHub and executed in CI. That matches how developers already build and debug features, and avoids forcing teams into a separate cloud testing surface for everyday authoring.
-
GitHub and Microsoft are clearly pushing testing into the IDE and planning layer. Copilot added test generation in VS Code in September 2024, agent mode with MCP support reached VS Code stable in April 2025, and Azure Boards can now create draft pull requests with Copilot to improve test coverage and other tasks.
-
The competitive split is becoming clearer. Cypress keeps winning with open source adoption and deep debugging, while QA Wolf sells a managed service with humans in the loop. Momentic sits in the middle as developer owned infrastructure, aiming to replace brittle scripts without outsourcing the workflow.
This category is heading toward workflow control as the main battleground. The products that win will combine IDE presence, pull request gating, local execution, and eventually synthetic monitoring into one system, so the same test definition can help write code, catch regressions before merge, and watch critical paths after deploy.