Vibium: What We're Building and Why | Jason Huggins

Описание: Video courtesy of TestingBot, our conference video sponsor. Run your tests in the cloud: https://testingbot.com/features.

Vibium: What We're Building and Why | Jason Huggins

In agentic development loops, agents write code, drive browsers and phones, and verify their own work. Test automation has to fit inside that loop: fast enough to keep up, structured enough for an agent to reason about, and able to answer questions assertions weren't designed for, like what does good look like?, without a human in the loop.

That's the problem behind Vibium, an open source, AI-native test automation stack built on WebDriver BiDi, spanning device setup, local execution, sandboxing, orchestration, federation across teams, and a hosted cloud on top. This talk walks through what's being built and why: the architectural bets, current state, and how to get involved.

Outline

1. Opening. Twenty years of clicking buttons. Selenium, Appium, the Tapster robot side quest, and what changed when agents showed up.
2. Why. The question agents can't answer with assertions: what does good look like? Why WebDriver BiDi is the right foundation, why model-based testing finally works now, why pass/fail isn't enough.
3. What. The multiple parts of Vibium and how they fit together: device setup, local execution, sandboxing, orchestration, federation, and hosted cloud. Session recordings as the cross-cutting layer, and why the format needs to be open.
4. Where it's headed. What's ready today and what's next.
How to plug in. For users, contributors, and other tool authors.
5. Q&A.

Takeaways
1. Why agentic development loops change what test automation needs to do.
2. What "good" means when there's no human in the loop, and why assertions aren't enough.
3. The shape of an AI-native test automation stack, and the bets behind it.
4. How to get involved.