Retrieval-augmented generation (RAG)

RAG is a way of making an AI assistant answer from your own approved documents. It looks up the relevant passages first, then writes the answer from them.

Why it matters

It cuts made-up answers and lets every reply point to its source. It's how a support bot or sales assistant stays accurate about your business.

In practice

  • Start with the documents you'd trust a new hire to read.
  • Keep them current and remove the outdated ones.
  • Show the source with every answer.

Context systemsAll 42 terms

Related terms

Structured data

Structured data is machine-readable markup, usually JSON-LD, that tells search engines and AI tools exactly what a page is about: your organisation, services, people, prices, FAQs.

Learn the term

Technical content

Technical content explains products and methods precisely: specifications, standards, tolerances and how things are made or tested.

Learn the term

The two readers

Every website now has two readers: the person deciding in seconds, and the AI tool they asked first.

Learn the term

Want this done for you?

Retrieval-augmented generation is part of a normal project here. Tell me where you are and I'll suggest the first three fixes.

Book a 20-minute call