AI agent system
Docs RAG That Cites
Answers from your documentation, with the paragraph it used
by Nadia Okafor
What it does
A retrieval system over your own documentation that answers in prose and shows the source paragraph for every claim. If it cannot find support, it says so instead of composing something plausible.
Ships as a container with a pgvector store. Re-indexes on a webhook when your docs change, so the answer and the documentation never drift apart.
Ideal use cases
- Internal knowledge bases above a few hundred pages
- Public docs where a wrong answer costs a support ticket
- Teams that need every answer traceable to a source
Where not to use it
- Documentation that is out of date — this makes that worse, not better
- Questions whose answers are not written down anywhere
The problem it solves
Support and engineering answer the same documented questions all week, and the answers drift from what the docs actually say.
Details
- Domain
- Category
- Sub-category
- Architecture
- Built for
- Runs on
- Protocols
- Ecosystems
- Deployment