Why my clinical RAG is mostly SQL
In the previous post I built a RAG system over my One Piece notes. At the end, I said I'd tackle something more real next: RAG for the healthcare industry. So here we are.
The anime version was fun, a
codeanding.com8 min read
Hayrullah Kar
Founder & Automation Strategist @MageSheet | Google Apps Script & AI Specialist.
Splitting structured clinical facts into SQL and freeform text into pgvector is pure architectural maturity. Relying on evals rather than unit tests for confident wrong answers is spot on.