← Presentation home · Part 2 of 4
Problem we are solving
- Clinic I work in: policies and procedures, answers needed now
- A stock internet AI still answers, and it makes mistakes, because it does not have our docs
- RAG (retrieval-augmented generation) can reduce those mistakes when it retrieves the right sources. Answers and citations still need checking
- You can do this RAG stuff in Claude. It can work, but the documents are processed by a third-party service rather than only on your own hardware
- We wanted the model and the files on hardware we own