> ## Documentation Index
> Fetch the complete documentation index at: https://docs.famulor.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Knowledge Bases (RAG)

> Teach your assistant with websites, PDFs, text files, and docs so it can answer business-specific questions reliably.

With **Knowledge Bases (RAG)**, you can provide your assistant with company-specific information such as pricing, opening hours, policies, and product details.

Supported sources include:

* Websites
* PDF files
* Text files
* Docs files

<div style={{ position: "relative", paddingBottom: "56.25%", height: 0 }}>
  <iframe style={{ position: "absolute", top: 0, left: 0, width: "100%", height: "100%", border: 0 }} src="https://www.tella.tv/video/vid_cmn32kr1w02kl08ml8n5u14f6/embed?b=0&title=0&a=1&loop=0&t=0&muted=0&wt=0&o=1" allowFullScreen allowTransparency />
</div>

## 1. Create a knowledge base

1. Open **Knowledge Base** in the left menu.
2. Click **New knowledge base**.
3. Add a name and description.
4. Click **Create**.

Your new knowledge base will then appear in the overview.

## 2. Add documents

Inside the knowledge base, use **Create a new document** to import content.

A common starting approach is a **website**:

1. Paste the URL.
2. Use **Load links automatically** to include subpages.
3. Increase rate limit if needed for larger sites.
4. Create the document.

You can also upload files via **drag and drop**, then name and describe them.

## 3. Performance and latency

Important: very large knowledge bases can increase latency.

If you have many heavy PDFs, converting core content to **text files** often improves processing speed.

## 4. Connect it to your assistant

In your assistant, open **Prompt & Tools** and go to **Knowledge Base**:

1. Select the target knowledge base.
2. Enable **Function calling** mode.

After that, your assistant can retrieve the stored information live during calls.

## 5. Test it

Use **Test Assistant** and ask targeted questions (for example pricing or opening hours) to verify answer quality and speed.

<Tip>
  Start with your highest-impact information first, then expand gradually to keep quality high and latency stable.

  Related docs: [Knowledge bases](/en/conversation-design/knowledge-bases), [Prompt & tools](/en/ai-assistants/prompt-and-tools), and [Webhook API](/en/automation/webhook-api).
</Tip>