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

# Knowledge bases

> The workspace that ties your documents, data, prompts, and agents together.

A **knowledge base** is the workspace where everything comes together. It holds the documents your assistant answers from, the database connections it can query, the prompts that shape its voice, the agents built on top of it, and the chats that run against it. Most of your day-to-day configuration happens here.

Every knowledge base belongs to exactly one [tenant](/product/concepts), and retrieval stays within its private corpus. [Agent sharing](/knowledge-bases/agent-sharing) passes another agent's final answer to the caller without exposing or merging the other knowledge base's documents.

## What a knowledge base contains

<CardGroup cols={2}>
  <Card title="Documents" icon="file-lines" href="/knowledge-bases/documents">
    The searchable corpus. Upload files, ingest a URL, or crawl a website, and the assistant answers from grounded, cited passages.
  </Card>

  <Card title="Prompts" icon="pen-nib" href="/knowledge-bases/prompts">
    Shape who the assistant is and how it answers with a persona, instructions, and optional full custom prompts.
  </Card>

  <Card title="Memory modes" icon="brain" href="/knowledge-bases/memory-modes">
    Decide whether and how this knowledge base remembers durable facts across conversations.
  </Card>

  <Card title="Agent sharing" icon="diagram-project" href="/knowledge-bases/agent-sharing">
    Let agents in other knowledge bases delegate to this one — while keeping your documents private.
  </Card>
</CardGroup>

## Anatomy

<AccordionGroup>
  <Accordion title="Name and description" icon="tag">
    A human-friendly name and description that identify the knowledge base in the dashboard and help your team tell them apart. The description is a good place to note what the corpus covers and who it is for.
  </Accordion>

  <Accordion title="Status" icon="signal">
    A knowledge base can be **active**, **inactive**, **pending**, or **suspended**. An active knowledge base is ready to answer questions; the other states let you stage or pause a knowledge base without deleting its content.
  </Accordion>

  <Accordion title="Documents" icon="file-lines">
    The corpus the assistant retrieves from. Documents are processed in the background and become searchable once they reach the **processed** state. See [Documents & ingestion](/knowledge-bases/documents).
  </Accordion>

  <Accordion title="Database connections" icon="database">
    Optional read connections that let the assistant query your live data and chart the results, alongside answers grounded in your documents.
  </Accordion>

  <Accordion title="Prompts" icon="pen-nib">
    The persona, instructions, and optional custom prompts that define the assistant's voice and rules. See [Prompts](/knowledge-bases/prompts).
  </Accordion>

  <Accordion title="Agents" icon="robot">
    Saved, deployable configurations that run on this knowledge base. See [Agents](/agents/overview).
  </Accordion>
</AccordionGroup>

## Quotas

Each tenant comes with quotas that keep usage predictable:

<ParamField path="max knowledge bases" type="number">
  The most knowledge bases the tenant can hold.
</ParamField>

<ParamField path="max documents" type="number">
  The most documents the tenant can store across all of its knowledge bases.
</ParamField>

<ParamField path="max storage" type="number">
  The total storage the tenant's documents can occupy.
</ParamField>

<Note>
  When a new knowledge base or document would push a tenant past one of its quotas, the request is refused with a clear message telling you which limit was reached. Remove content you no longer need, or raise the tenant's quota, and try again.
</Note>

## Create a knowledge base

<Steps>
  <Step title="Name it">
    Give the knowledge base a clear name and a short description of what it covers.
  </Step>

  <Step title="Add content">
    Upload documents, ingest a URL or website, and optionally add a database connection. See [Documents & ingestion](/knowledge-bases/documents).
  </Step>

  <Step title="Shape the voice">
    Set a persona and instructions so the assistant answers in the tone and within the rules you want. See [Prompts](/knowledge-bases/prompts).
  </Step>

  <Step title="Start chatting or build an agent">
    Ask questions directly, or save the configuration as an [agent](/agents/overview) you can deploy, schedule, or call through the API.
  </Step>
</Steps>

## Next steps

<CardGroup cols={2}>
  <Card title="Add documents" icon="file-arrow-up" href="/knowledge-bases/documents">
    Upload files, ingest a URL, or crawl a website.
  </Card>

  <Card title="Shape the assistant" icon="pen-nib" href="/knowledge-bases/prompts">
    Set a persona, instructions, and custom prompts.
  </Card>
</CardGroup>