The interface is genuinely good, and the Thunders copilot is what makes the whole thing click. It gives you confidence while you are building a test, and it visibly raises the quality of what you end up writing rather than just getting you to something that runs.<br><br>The product team ran the evaluation, but the effect went well beyond us: it raised the maturity of the whole team, product and engineering alike, on testing as a subject. Because writing a test no longer requires being technical, it became a shared conversation instead of a specialist one.<br><br>Concretely, it let us redraw roles and responsibilities. Product now owns very precise acceptance tests and edge case QA, where our knowledge of the expected behaviour is the differentiator. Engineering keeps the rest. That split was not realistic before, and it is the change I value most.<br><br>The practical result is that we write end to end non regression tests much earlier in the cycle. Previously that was a project in itself, so it kept being postponed. We now go into releases with a high level of confidence that existing behaviour has not broken.

What stood out most during our trial was generating tests in natural language. As a growing enterprise B2B SaaS startup, we have multiple products, languages, and customer-specific requirements, but no dedicated automation engineers yet. Making automation accessible to our existing QA team is particularly valuable. From the trial, demo, and conversations with the team, I see strong potential to expand our test coverage and reduce the manual regression testing bottleneck before releases.

What I like most is the simplicity of the approach. You can describe what you want to test in plain language and let the AI agent handle the browser, instead of spending time maintaining fragile automation scripts. I’m not a daily production user, but from what I’ve seen, this could be especially useful for manual QA teams that want to automate more without adding a lot of technical overhead.