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

# Knowledge Base

> Upload documents, scrape websites, and add custom content so your AI agents can answer questions accurately during calls.

## What is a Knowledge Base?

A Knowledge Base is a collection of documents and content that your AI agents can reference during calls. Instead of relying only on their system prompt, agents search your knowledge base in real-time to give accurate, specific answers about your business.

## How retrieval works

When a caller asks a question, the agent searches your knowledge base for the most relevant passages and answers from those, not from guesswork.

```mermaid theme={null}
flowchart LR
    Q(["Caller asks a question"]) --> S["Agent searches<br/>the knowledge base"]
    subgraph KB["Your knowledge base"]
      D1["Uploaded files"]
      D2["Scraped pages"]
      D3["Custom answers"]
    end
    S --> KB
    KB --> R["Top matching passages"]
    R --> A(["Agent answers, grounded in your content"])
```

## Adding Content

Navigate to **Agent Studio** and select your agent. In the **Knowledge Base** section, you have four ways to add content:

### Upload Files

Drag and drop files or click to browse. SmartAlex processes them automatically.

| File Type    | Extensions  | Processing         |
| ------------ | ----------- | ------------------ |
| **PDF**      | .pdf        | Parsed and chunked |
| **Word**     | .doc, .docx | Parsed and chunked |
| **Text**     | .txt        | Ready immediately  |
| **Markdown** | .md         | Ready immediately  |
| **CSV**      | .csv        | Ready immediately  |

**Limits:** Up to 10 files at a time, 20MB max per file.

### Scrape Website (single page)

Enter a single URL and SmartAlex will extract the content from that page:

1. Click **Scrape Website**
2. Enter the full URL (e.g., `https://example.com/about`)
3. SmartAlex fetches and processes the page content
4. A new document appears in your knowledge base

### Sitemap Scrape (multiple pages)

Extract content from multiple pages at once using your sitemap:

<Steps>
  <Step title="Enter your domain">
    Type your website domain (e.g., `example.com`) and click **Load**.
  </Step>

  <Step title="Select pages">
    SmartAlex discovers all pages from your sitemap. Select which pages to scrape (up to 50 per batch).
  </Step>

  <Step title="Start scraping">
    Click **Scrape N Pages**. Each page becomes a separate document in your knowledge base.
  </Step>
</Steps>

### Write Manually

Create a document from scratch:

1. Click **Write Document**
2. Enter a title (e.g., "Return Policy", "Office Hours")
3. Write or paste your content
4. Click **Create**

## Document Processing

After adding content, documents go through automatic processing:

| Status         | Meaning                                         |
| -------------- | ----------------------------------------------- |
| **Pending**    | Queued for processing                           |
| **Processing** | Being split into chunks and indexed             |
| **Completed**  | Available to your agent (shows chunk count)     |
| **Failed**     | Processing error, retry or delete and re-upload |

Text files (.txt, .md, .csv) are ready immediately. PDFs and Word documents require processing time.

## Connecting Documents to Agents

Each document can be connected to one or more agents:

1. Open your agent in **Agent Studio**
2. Go to the **Knowledge Base** section
3. Toggle the checkbox next to each document you want the agent to use
4. Connected documents show a checkmark

<Note>
  Connected documents are available to your agent during calls. The agent searches relevant content to answer questions accurately.
</Note>

## How Agents Use Knowledge

During a call, when a caller asks a question:

1. The agent identifies the question topic
2. It searches connected knowledge base documents
3. The most relevant content chunks are retrieved
4. The agent uses this context to give an accurate answer

This happens in real-time, callers won't notice any delay.

## Managing Documents

| Action         | How                                                                                |
| -------------- | ---------------------------------------------------------------------------------- |
| **Preview**    | Click the eye icon to view full content and metadata                               |
| **Search**     | Use the search bar to filter documents by title                                    |
| **Delete**     | Click the trash icon, this is permanent                                            |
| **Disconnect** | Uncheck the document to stop the agent from using it (doesn't delete the document) |

## Best Practices

* **Keep documents focused**, one topic per document works better than one large document
* **Update regularly**, delete and re-upload documents when your business information changes
* **Start with your FAQ**, the questions your receptionist answers most often are the best starting point
* **Test after adding**, call your agent and ask questions to verify it finds the right answers
* **Use descriptive titles**, helps you organise and find documents later