# Bring your business knowledge into the conversation.

Product: RAG

Give AI agents relevant information from your services, policies and FAQs. Maintain the source content your agents retrieve when customers ask questions.

Interactive product examples use illustrative data; changes remain on the page.

## Answer a service-policy question from the source.

Retrieval-augmented generation brings relevant passages into the agent's answer context, so it can work with your business information.

### Add the information customers need

Create focused sources from supported text, FAQs, files or website content. Let processing complete before using them in a conversation.

### Connect knowledge to its role

Assign the knowledge base to the relevant assistant. Give a service agent the policies and offering details appropriate to its job.

### Retrieve the relevant detail

When asked a question, the assistant can use selected source passages alongside its instructions. Questions requiring live information can use a configured tool.

## Give the agent context your business owns.

### Business detail beyond general knowledge

Bring your actual services, policies and common answers into the conversation without training a new model.

### A maintained place for guidance

Update the source material when an offer or policy changes, then process it for the assistants assigned to that knowledge.

### Information matched to the task

Use retrieval for maintained guidance and connected APIs for live calendars, orders or authorized account information.

## Keep the sources as deliberate as the answers.

Useful retrieval depends on the information you provide, the questions you test and the response when an answer is missing.

### Clear source ownership

Maintain approved content and remove conflicting or outdated guidance. URL imports need updating when the original website changes.

### A path for missing information

Instruct the assistant to clarify or involve your team when its sources do not answer the question. Retrieval does not guarantee every answer is correct.

### Relevant conversation checks

Test representative customer questions on the intended channel and revisit affected questions after changing a source or assistant assignment.

## Product capabilities and architecture

Assistable RAG, or retrieval-augmented generation, supplies relevant passages from assigned knowledge sources when an AI agent answers. Add text, FAQs, website content or files to give the assistant business-specific context.

### From business content to a relevant answer.

### Text, FAQs, URLs and files

Build a knowledge base from supported source types. Use focused content that reflects what customers actually ask about your business.

### Relevant passage retrieval

Bring selected source material into the model's answer context. The assistant can use business detail beyond its general knowledge.

### Assistant-level assignment

Attach the processed knowledge to the assistants that need it. Keep a service agent focused on the information appropriate to its role.

### Maintained source content

Update the underlying information when services, policies or offers change, then process the source for the agent to use.

### Business context at the point of the answer.

Use maintained content for guidance and connected tools for facts that change.

### Sources provide the reference

Retrieval brings your approved content into the answer without retraining the underlying model.

### Instructions shape the response

Define how the assistant explains information, clarifies requests and handles gaps in the sources.

### Tools supply changing facts

Use connected APIs for current availability, order status and account records.

## Operating controls and references

### Useful answers start with accountable sources.

Keep ownership of the information as clear as ownership of the agent.

### Approved content

Give each source an owner and remove outdated or conflicting guidance. Attach only the information appropriate to the assistant's role.

### A defined response to missing information

Instruct the agent to clarify or escalate when its sources do not answer the question, rather than filling gaps with an unsupported business claim.

### Controlled updates

Review source processing and affected questions after policy changes. URL imports need maintenance when the original website changes.

- [Add and maintain knowledge sources](https://docs.assistable.ai/v3/guides/knowledge/add-sources)
- [How knowledge retrieval works](https://docs.assistable.ai/v3/guides/knowledge/retrieval)
- [Connect a live-data tool](https://docs.assistable.ai/v3/guides/tools/custom-api)

## Connect the next part of your workflow.

- [Chat AI](https://www.assistable.ai/solutions/chat-ai): Qualify, book and support through customer messaging.

- [Tooling](https://www.assistable.ai/solutions/tooling): Connect conversations to calendars, records and APIs.

- [Observations](https://www.assistable.ai/solutions/observations): Know which conversations need attention.

## RAG questions.

### What does RAG mean for an AI agent?

Retrieval-augmented generation adds relevant business information to the context used for an answer. It lets an assistant work with maintained services, policies and FAQs without training a new model.

### Does a knowledge base prevent every incorrect answer?

No. Source quality, retrieval and interpretation all affect the response. Maintain clear information, define the behavior for missing answers and evaluate representative customer questions.

### Should live availability and account records go in the knowledge base?

Use a connected tool when information changes frequently or requires authorization. RAG suits maintained guidance; calendar, order and account APIs suit live operational facts.

### Are imported website pages continuously synchronized?

A URL import is not a continuous synchronization guarantee. Update and process the source when the underlying content changes so the assistant has the current information.

## Put your best business knowledge to work.

Bring the questions customers ask and the sources your team trusts. We will help connect them to a useful agent experience.

[Talk to sales](https://www.assistable.ai/contact)

Source: https://www.assistable.ai/solutions/rag