I have added a lot of new features and fixed quite a few bugs in GExperts during the last weeks. I have been so productive because I extensively used Claude Code, and it is a lot of fun.
On the other hand: Does it really make sense to invest time, effort and of course money (The max plan costs 100 Eur/month.) into something that more and more feels like a legacy tool (GExperts) for another legacy tool (the Delphi IDE)? I have rarely started Delphi for about 2 Months! Most of my work was done with and by Claude Code from the command line and using only the command line compiler. The generated code isn’t really worse than the one I would have written myself, sometimes it is a lot better. Also, since I now always start new features with unit tests before the implementation, because I don’t have to write those boring tests myself, the test coverage is much better than it used to be.
I really wonder about that. Maybe it would be a better use of my time to learn even more about using AI and how to integrate it better into my workflow. I have been a skeptic of LLMs for a long time but no longer. I’m a fan now. If you are a software developer and haven’t done it yet, I urge you to invest 20 Euro/month into a Pro plan and try Claude Code yourself. It’s an eye opener!
The worst part: I still don’t understand how Claude Code can be so powerful. From what I think I know about how LLMs work, they should not be able to do what it does.
Discussion about this in the corresponding post in the international Delphi Praxis forum.