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

# Building agents

> Create, configure, and deploy AI agents across phone, WhatsApp, Web Talk, and Web Chat.

An **agent** is the AI that talks to your customers. You build and configure it in the
**Agent Studio**, where you define how it behaves, what it knows, what actions it can
take, and which channels it uses.

## Create an agent

You can start an agent two ways:

<CardGroup cols={2}>
  <Card title="From a template" icon="template">
    Begin with a ready-made starting point — the name, prompt, language, and voice are
    pre-filled so you can adapt rather than start from a blank page.
  </Card>

  <Card title="From scratch" icon="pencil">
    Start empty with just a name, a short description, and your own prompt, then configure
    everything yourself.
  </Card>
</CardGroup>

Once created, the agent opens in the Agent Studio, ready to configure.

## How the Agent Studio is laid out

The Studio has two parts that are always working together:

* **Agent Core** — a panel that stays visible while you work. It holds the essentials:
  the agent's **persona and behavior**, its **greeting**, its **Agent Purpose**, **Response
  Style**, **Call Flow**, and [variables and default values](/building-agents/variables).
* **Configuration tabs** — everything else, grouped into focused sections you move between.
  They appear in this order: **Intelligence**, **Capabilities**, **Voice**, **Runtime**,
  **Outcome**, and **Deploy**.

<CardGroup cols={2}>
  <Card title="Agent Core" icon="id-badge" href="/building-agents/agent-core">
    Persona, behavior, greeting, prompt sections, and [variables and default values](/building-agents/variables).
  </Card>

  <Card title="Intelligence" icon="brain" href="/building-agents/intelligence">
    Choose the language model and tune Temperature and Max Tokens. Lower temperature keeps
    responses focused; higher temperature makes them more creative. Max Tokens sets the maximum
    response length per turn.
  </Card>

  <Card title="Capabilities" icon="plug" href="/building-agents/capabilities">
    Set up the three capability areas: Knowledge Base, Integrations, and Custom APIs.
  </Card>

  <Card title="Voice" icon="wave-sine" href="/building-agents/voice">
    Choose a TTS provider and voice, adjust Speaking Speed, and preview the voice in
    Browser (HD) or Phone (16kHz) quality.
  </Card>

  <Card title="Runtime" icon="adjustments" href="/building-agents/runtime">
    Configure Settings for conversation behavior, natural closing, duration limits, and
    silence handling, plus Transfer for human handoff, transfer numbers, and caller messages.
  </Card>

  <Card title="Outcome" icon="clipboard-check" href="/building-agents/outcome">
    Configure recordings, transcripts, AI Analysis, and Custom Analysis after each conversation.
  </Card>

  <Card title="Deploy" icon="rocket" href="/building-agents/deploy">
    Use Routing for inbound and outbound numbers, Widget for Web Talk or Web Chat setup,
    and WhatsApp to assign and configure a WhatsApp number.
  </Card>
</CardGroup>

## One agent, multiple channels

Each agent can work across multiple channels: phone calls, WhatsApp, Web Talk, and Web
Chat. You configure the agent once, then use Deploy to connect and tune the channel
settings it needs.

<Note>
  Designing for WhatsApp is different from designing for a phone call — message length,
  templates, and the 24-hour window all shape the experience. Vocily keeps the same agent
  model, while [Deploy](/building-agents/deploy) lets you configure the channel-specific
  setup for each place the agent runs.
</Note>

## Saving and going live

Changes take effect when you **save** — there's no separate publish step. Once an agent is
connected to a channel under [Deploy](/building-agents/deploy), your saved configuration is
what handles live conversations.

## Test before you rely on it

At any point you can [test the agent](/test/overview) directly in the app — by chat, by
voice in your browser, or by placing a real phone call — before you send real customers to
it.

<Card title="Start with Agent Core" icon="arrow-right" href="/building-agents/agent-core">
  The persona, prompt, and language are the foundation of every agent.
</Card>