My workflow is very task-oriented. I often have one AI generate code that another reviews and corrects, which lets me jump into entirely new areas quickly. I’ve also experimented with fully automated, idea-to-production platforms (like v0, Replit, and a few others). However, my experience with those has been mixed—they often end up requiring more manual effort than their vision-driven claims would suggest (though I expect they’ll catch up soon).
So, I’m curious: Is it just me, or are these new AI-powered IDEs truly effective? Are you using them exclusively for new projects or also for maintaining existing codebases? And how large are the projects you’re managing to complete fully automated?
I'm seeing lots of hype, and some real amazing results, but all of the new tools still seem to struggle at being a net positive when codebases get large or non-trivial. Makes it much much easier to build something useful but basic, but not that much easier to build something large or excellent.