> ## 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.

# Chat overview

> How AutoSage turns a question into a grounded, cited answer against your knowledge base.

A **chat** is a conversation against a knowledge base. When you send a message, AutoSage doesn't answer from thin air — it can search your documents, query connected databases and chart the results, use connected apps and tools, and cite supporting sources. Streaming responses can show live progress before delivering the final answer.

The same chat engine powers both dashboard conversations and [agent runs](/agents/overview), so what you learn here applies everywhere in AutoSage.

## What a chat can do

<CardGroup cols={2}>
  <Card title="Search your documents" icon="magnifying-glass">
    Finds the most relevant passages in the knowledge base and cites them in the answer.
  </Card>

  <Card title="Query connected data" icon="database">
    Runs read queries against connected databases and can chart the results inline.
  </Card>

  <Card title="Use tools and apps" icon="plug">
    Calls connected apps, custom tools, and other agents to go beyond the documents.
  </Card>

  <Card title="Cite its sources" icon="quote-left">
    Shows where each part of the answer came from, so you can verify it.
  </Card>
</CardGroup>

## How a chat is organized

Every chat belongs to a **tenant** and a **knowledge base**, and each one carries a **privacy level** that controls who can see it.

<Steps>
  <Step title="Tenant">
    The resource owner the chat runs under — its credits and quotas apply to the conversation.
  </Step>

  <Step title="Knowledge base">
    The content the chat is grounded in: documents, database connections, prompts, and tools.
  </Step>

  <Step title="Privacy level">
    Private, Organization, or Public — determines who can view the conversation. See [Sharing a chat](/chat/sharing).
  </Step>
</Steps>

<Info>
  Because a chat is scoped to one knowledge base and tenant, its answers only ever draw on that knowledge base's content. This keeps every conversation grounded in exactly the material you intend.
</Info>

## How a question gets answered

<Steps>
  <Step title="Understand the message">
    AutoSage reads your question along with the recent conversation so far.
  </Step>

  <Step title="Gather grounded context">
    It searches your documents, and when helpful runs data queries, uses tools, or recalls durable [memory](/memory/overview).
  </Step>

  <Step title="Compose the answer">
    The model writes a grounded response. Streaming shows live progress and delivers citations and final content as they become available.
  </Step>
</Steps>

## Choosing how deeply to answer

Each message can be answered in one of two modes — a fast single-pass **Quick mode** or a thorough multi-step **Deep mode**. Pick the right one per message to balance speed and depth.

<Card title="Compare the modes" icon="gauge-high" href="/chat/modes">
  See when to use Quick versus Deep answering.
</Card>

## Next steps

<CardGroup cols={2}>
  <Card title="Quick vs Deep modes" icon="gauge-high" href="/chat/modes">
    Choose the right answering mode for each message.
  </Card>

  <Card title="Streaming responses" icon="bolt" href="/chat/streaming">
    See how live responses arrive and what to expect.
  </Card>

  <Card title="Sharing a chat" icon="share-nodes" href="/chat/sharing">
    Share a read-only transcript with your org or the world.
  </Card>

  <Card title="Long conversations" icon="clock-rotate-left" href="/chat/long-conversations">
    How AutoSage keeps lengthy chats fast and coherent.
  </Card>
</CardGroup>