# AI Unit Tests: Hype vs My 4-Hour Night

AI tools promise: "Unit tests in seconds!" (Copilot, Claude, etc.)

My night struggling with Aider and local (small) LLMs:

- AI invented "ApiKeys" table (real name: Authorizations).
- Picked wrong test tool (MSTest instead of xUnit).
- Said "Done!" but files empty.

Still no good results in 4 hours (I could have done this manually in 1 hour).
Lesson: AI helps a lot, but you must check everything and take a lot of time to prepare.

With my technical background, I prefer to explore and gain more experience rather than buying off-the-shelf cloud products. Free local AI (like Phi3, Qwen2.5, DeepSeek) = unlimited power; cloud AI is smoother but expensive and less secure.

(Deja vu: Same as with my multi agent hangman game: days of tuning prompts/history/temp!)

Anyone else trying Aider with local LLMs? Which model?

---

*Co-authored with Claude.*

If this was useful, [you can buy me a coffee](https://ko-fi.com/p47k0).

New articles: [follow via RSS](https://blog.koorevaar.com/feed.xml).