Retrieval-augmented generation

A language model alone answers from what it absorbed in training, which can be out of date or simply wrong about your business. With retrieval-augmented generation the system searches an approved source (your price list, policies, service manuals, the caller's account) and hands the results to the model with instructions to answer from them and to say so when the answer is not there.

This is how an AI voice agent knows your hours, your service area and your cancellation policy without being retrained, and how a support assistant quotes the right manual. For a growing business it also means the content you maintain is the content the AI uses, so keeping documents current becomes an operations task.

Ask a vendor what sources the system retrieves from, how often they are refreshed, whether it cites or logs the source it used, what it says when nothing relevant is found, and how you correct a wrong answer. A demo on your own documents is worth more than any benchmark.