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

# Configure your XUNA AI Conversational AI agents

> Explore every configuration option for your voice agent: system prompts, LLM selection, voices, knowledge bases, tools, and authentication.

Every agent you build on XUNA AI Conversational AI is defined by a set of configuration layers — from the words it speaks to the APIs it can call. This section walks you through each layer so you can build agents that are accurate, expressive, and secure.

<CardGroup cols={2}>
  <Card title="System prompt" icon="file-lines" href="/configure/system-prompt">
    Write the instructions that shape your agent's personality, behavior, and goals.
  </Card>

  <Card title="Models" icon="cpu" href="/configure/models">
    Choose from GPT, Gemini, Claude, or a custom LLM. Tune temperature, reasoning, and fallback behavior.
  </Card>

  <Card title="Voice and language" icon="microphone" href="/configure/voice-language">
    Pick from 5,000+ voices across 31 languages and control speech style.
  </Card>

  <Card title="Knowledge base" icon="book" href="/configure/knowledge-base">
    Ground responses in your own documents, URLs, or text snippets using retrieval-augmented generation.
  </Card>

  <Card title="Tools" icon="wrench" href="/configure/tools">
    Connect your agent to external APIs, client-side actions, MCP servers, and built-in system tools.
  </Card>

  <Card title="Conversation flow" icon="arrows-split-up-and-left" href="/configure/conversation-flow">
    Control turn-taking, interruption handling, and timeout behavior.
  </Card>

  <Card title="Personalization" icon="user-gear" href="/configure/personalization">
    Inject per-session variables and override voice, prompt, or language at runtime.
  </Card>

  <Card title="Authentication" icon="lock" href="/configure/authentication">
    Secure agent access with signed URLs, conversation tokens, or custom middleware.
  </Card>
</CardGroup>

## Configuration layers

Each layer builds on the one before it. You can configure most options in the dashboard or via the API — both are equivalent.

<Steps>
  <Step title="Define behavior">
    Write a [system prompt](/configure/system-prompt) that tells the agent who it is and what it should do.
  </Step>

  <Step title="Choose a model">
    Select an [LLM](/configure/models) that fits your latency, quality, and compliance requirements.
  </Step>

  <Step title="Set the voice">
    Assign a [voice and language](/configure/voice-language) to match the agent's persona.
  </Step>

  <Step title="Add knowledge">
    Upload files or point to URLs so the agent can answer questions grounded in your content using the [knowledge base](/configure/knowledge-base).
  </Step>

  <Step title="Connect tools">
    Give the agent the ability to take action by connecting [tools](/configure/tools) — from API calls to transferring calls.
  </Step>

  <Step title="Tune conversation flow">
    Adjust how the agent handles [turn-taking and interruptions](/configure/conversation-flow) to feel natural on your platform.
  </Step>

  <Step title="Personalize per session">
    Pass runtime data like user names or account details using [dynamic variables](/configure/personalization).
  </Step>

  <Step title="Secure access">
    Gate who can start conversations using [authentication](/configure/authentication).
  </Step>
</Steps>

<Note>
  All configuration options are also available via the [REST API](/api-reference/overview). Use the API to manage agents programmatically or to automate deployments.
</Note>