InfiVoiceFeaturesPricingBlog

Knowledge base (RAG) for voice agents: what it is and how an AI voice agent uses it

InfiVoice ยท 2026-11-14

A knowledge base with RAG (retrieval-augmented generation) is the set of your documents that the agent searches before it answers. That way the agent replies with your real information, not guesses.

What RAG is

RAG stands for retrieval-augmented generation. When someone asks a question, the system searches your content for the relevant passages and gives them to the model, which writes the answer from those passages.

How an AI voice agent uses it

During the call, the agent:

1. Understands the question in natural language.

2. Searches your knowledge base for the passages that answer it.

3. Answers out loud with that information and continues the flow you designed, for example to book or transfer the call.

In InfiVoice you load your content into each agent's knowledge base and connect it to the flow in the visual builder.

What to load

If a detail is not in the knowledge base, the agent should not invent it. Design the flow to transfer the call or take a message.

RAG vs. a fixed script

Signs your knowledge base needs work

To prepare the content step by step, read the guide how to prepare a knowledge base for an AI voice agent.

Pricing

InfiVoice has a Free plan with 100 credits and a Pro plan at $79 per month with 250 credits. One credit equals 60 seconds of call time, so you can test your knowledge base on real calls before you pay.

Frequently asked questions

Is RAG the same as training the model?

No. RAG checks your documents on every answer. It does not change the model, so you can update information without retraining anything.

Can the agent answer in other languages using my knowledge base?

Yes. InfiVoice supports more than 30 languages, so the agent replies in the caller's language using your information.

How much does it cost?

Free includes 100 credits. Pro costs $79 per month with 250 credits; 1 credit is 60 seconds of call time.

Start free