I am huge fan of Google and the Google stack, but even though I get on well Gemini, they have felt constantly behind in the AI race. For a company that was setting the standard in areas from Quantum computing to Go it has been a little disappointing. Having tried Cursor and Claude Code as part of my professional workflow, experimenting with Antigravity leaves me thinking they are still not quite there. Although the autocomplete is good, I find the agentic features almost unusable. Yes, it is a free tool and yes in the current AI landscape where other companies are being accused of dumming down their tools and increasing their prices, this is a welcome change, but free or not, if the tool doesn’t deliver, I can’t use it.
I am plauged by timeouts and re-tries. It often takes multiple attempts to get any sort of response from the agentic features. Occasionally it gets stuck in a loop of tool selection thinking and loses track of files in my folder structure, even working in a very small project.
I want to like Antigravity. It feels like a polished VS Code, rather than Cursor’s more challenging changes to the IDE, so the switch has been fairly painless for the standard editor experience, but when the agentic AI features are so hit and miss, it just breaks the flow. I will keep trying for a while on small personal projects, but I can’t see myself using it for professional development any time soon.