Glossary

What is retrieval augmented generation (RAG)?

Retrieval augmented generation, or RAG, is a way of making a language model answer from your own documents. Before the model writes, the system searches a knowledge base for the passages most relevant to the question and gives them to the model, which then writes its answer from them. It reduces made up answers about your product.

Published by Sellinger AI. Numbers and settings checked on 24 October 2026 against the fact register. Report an error.

How does RAG work?

The steps of a RAG answer
Step	What happens
1. Store	Your documents are split into passages and indexed for search
2. Retrieve	A question comes in; the system finds the most relevant passages
3. Generate	The model writes an answer using those passages as its facts
4. Check	Good systems hand over when nothing relevant was found

Why does it matter for sales agents?

A sales agent is asked about your product all the time: what it costs, what it connects to, whether it works for a certain case. Without retrieval, the model answers from general knowledge and may invent. With it, the answer comes from what you wrote. The quality of the answer can only be as good as the knowledge base, so keep it accurate and current.

What makes a good knowledge base for RAG?

Short, specific passages: one topic each.
Plain facts: features, limits, prices you allow the agent to share, and what you do not do.
Answers to the questions buyers actually ask, taken from real replies.
Dates on anything that changes.

How Sellinger handles it

The agent answers from the knowledge base you write, together with your persona and rules. The AI knowledge base for sales agents entry and the knowledge base help article show how to set it up.

A campaign can run in approval mode, where every message waits for a person to approve it, or send automatically. A person can take over any conversation, and the agent stops on its own when it cannot answer, for example a price it has no answer for.

Common questions

Does RAG stop hallucinations?

It reduces them for questions your documents cover. For questions they do not cover, the agent should hand over.

Is RAG the same as training a model?

No. RAG gives the model documents at answer time. Training changes the model itself.

What should I put in a sales knowledge base?

Product facts, limits, approved prices, integrations, and answers to the questions buyers really ask.

Sources and method

General AI practice. Sellinger behaviour from the fact register, checked 24 October 2026.