
I work as a QA engineer and Gemini has become part of my daily routine. What I like most is handing it a user story or a set of acceptance criteria and getting back a decent list of test scenarios, including the negative and edge cases. It usually catches a boundary case or two that I'd have missed on my first pass. It's also pretty good with automation code. When a test fails with a long stack trace, I paste in the error along with the test and it normally points me in the right direction, like a selector that changed or a wait that's too short. Responses are quick, and because we already use Google Workspace, pulling in a requirements doc from Drive or exporting a table of test cases to Sheets is really convenient. Review collected by and hosted on G2.com.
It can be confidently wrong, which is a real risk in testing. A few times it gave me automation code that used a method that doesn't exist in the version we're on, or assumed a field would behave a certain way when it didn't. So I always double check before I use anything it gives me. In long conversations it also starts to lose track of details I mentioned earlier, like the exact validation rules for a form, and I end up repeating myself or starting a new chat. The answers can be a bit generic too unless I give it a lot of context upfront. It would be great if it held on to project context better between chats. Review collected by and hosted on G2.com.