Has anybody used one of the specialized products for AI coding?
I normally assume these things are just repackaged GPT calls that people resell… But I think I remember seeing an openAI video where they were talking about “partnering” with one or another of these companies. Just wondering what they actually do and if you have had good results.
I could imagine that collecting good library and api documentation and setting up some strong rag implementations would solve one of the biggest pain points I have with coding – when it gets that stuff wrong. So maybe the companies are doing that and it actually helps?
1 Like
I am created AI code editor with build in AI functions. I am using o1-mini. I also made code agent that perform as agent , debugging, refactoring, optimization, QA and testing.
1 Like
I’ve started using Codeium, but I’m still learning the ropes. 
Update, like most everyone else by now (?), I now use Cursor with Claude 3-7 and its quality varies depending on the task, but is sometimes super impressive. And certainly the best thing I’ve tried so far.
AI code assistance tools are already a dime a dozen. They are all excellent at finding your syntax errors, sometimes helpful at finding logical errors in very small code snippets, and virtually worthless for everything else. Personally I just ask the latest generative and/or reasoning models. Curiously, I find the generative models to often be superior to the reasoning models. Despite being advertised as better for coding, it seems their reasoning capabilities are often still weaker than predictive output.