To add to the point above: sounds like everyone is leaning towards Cursor or Cody. If youre really watching that $30 budget tho, definitely check out Codeium Individual Extension Free. Its actually amazing and wont cost you a dime! I switched for my Python scrapers and the logic is way more reliable. It handles those big CSV exports like a champ!

Re: > im honestly so fed up with Copilot... I totally get the frustration. You might want to consider Sourcegraph Cody AI Pro Subscription instead. It pulls context from your whole project way better. Huge for scrapers. Just be careful with how you define your schemas tho... it can still get overconfident. I would suggest using it with the OpenAI GPT-4o Multimodal LLM to handle those complex CSV exports so you dont miss that Friday deadline.

> im honestly so fed up with Copilot Cursor AI Code Editor with Anthropic Claude 3.5 Sonnet is a solid choice. It handles logic way better and hallucinates way less than Copilot in my exp.

I had a similar nightmare with a scraper project last year where I was pulling financial data and the AI kept making up pandas methods that didnt exist. It's a huge time sink. If youre worried about reliability and safety, especially with work-related data, I actually started using Tabnine Pro AI Coding Assistant because of how it handles privacy. Unlike some of the big cloud-first ones that just guess based on common internet patterns, Tabnine lets you prioritize your own local codebase which basically helps eliminate the imaginary library problem. It's not as flashy as some chat-based tools but the suggestions feel more grounded in technical reality. The reason it works for those massive CSV exports is the context management. When I was building a tool for a client in Denver, I needed to ensure it didnt use any GPL-licensed code by accident. This tool has a clean model training option where they only use permissively licensed code. It's a bit more boring but thats what you want when you have a Friday deadline. You dont want a creative AI; you want one that knows the Python standard library inside and out. Plus, it integrates well with VS Code without being too intrusive. It's worth a look if you need something that wont hallucinate weird dependencies just because it sounds right to a LLM. It keeps the scraper logic predictable... which is a lifesaver when you're dealing with Chicago-sized data sets.