As developers, we all end up writing SQL queries — sometimes simple, sometimes painfully complex.
But I kept asking myself:
👉 Why do I have to think in SQL when my brain is already thinking in English?
That’s what led me to build my latest side project: ChatWithDB, an MVP that lets you connect to your database and just… talk to it.
The Problem
SQL is powerful, but not everyone on a team is fluent in it:
- Product managers just want numbers.
- Analysts want insights without fighting syntax.
- Even developers sometimes waste time debugging queries instead of focusing on features.
And when you finally get the data, you often want to visualize it quickly.
My Approach
Instead of building a heavy “end-to-end AI analytics platform,” I wanted something light and fast:
- Connect your DB (Postgres/MySQL first).
- Ask in natural language → “Show me daily signups in the last 30 days.”
- Get SQL + results instantly.
No extra modeling. No complex setup. Just query → answer.
Lessons Learned While Building
- Speed matters: People lose trust if the model takes forever to respond.
- Schema awareness is tricky: LLMs need context, but too much context slows things down.
- UI simplicity wins: If it feels like ChatGPT, adoption is easier.
What’s Next
Right now, I’m testing whether a lighter approach (fast responses, no over-engineering) is more valuable than big enterprise-style systems.
I’d love feedback:
- Would you prefer a quick tool that handles 80% of your queries fast?
- Or a heavier system that tries to cover every possible edge case?
(If you’re curious, I’m hacking on this at [ChatWithDB](https://chatwithdb.com), still super early but feedback is welcome.)
Top comments (0)