Had high hopes and was very excited for this product. But it has proved time and time again… it’s not ready. The tool is, the Gemini AI is not. Wish I could use copilot or anthropic’s Claude.
The AI breaks my app more than it fixes. It has been trying to fix a single issue for over 24 hours now. It seems to get caught up in loops while trying to fix something (implements, then takes out, then re-implements, and again). The app has grown over the months (slowly) and the more complex the app, the worse the AI gets. The ai will even start to lie as it says it did something, but never does. Sometimes I have to tell it 3 times to implement the change.
A different app that I was trying to build, Gemini got so “lost” in trying to determine a NextJS issue that the site was no longer useable. Apparently I should commit my changes to github more often knowing Gemini will eventually break the app.