# Chat & Voice overview

> Configure conversational agents, knowledge, chat, phone resources, outreach, recordings, and analytics.

Canonical: https://help.spacebrain.ai/chat-and-voice/

Chat & Voice includes **Agents**, **Chat Widget**, **Knowledge**, outreach **Contacts** and **Campaigns**, **Calendar**, **SMS**, **Power Dialer**, **Phone Numbers**, **Recordings**, **Messaging Compliance**, **Analytics**, and reusable **Actions**.

## Recommended setup order

1. Define one agent purpose and its human owner.
2. Write boundaries, escalation rules, and prohibited claims.
3. Add a small, approved knowledge base.
4. Configure only the actions the agent needs.
5. Test the agent without a public number or widget.
6. Configure a chat widget or phone number and compliance prerequisites.
7. Run internal end-to-end tests.
8. Launch to a limited audience and review conversations, recordings, outcomes, cost, and handoffs.

## Design around a job, not a personality

A useful agent has a clear audience, opening, objective, information boundary, allowed actions, stop conditions, and handoff. Style and voice matter, but operational clarity matters more.

## Human accountability

Name the person who owns instructions, knowledge freshness, action permissions, quality review, and incident response. Every customer-facing agent should have a tested handoff path and a way to stop new traffic without deleting its configuration.

<Callout type="warn" title="Do not launch from a successful happy-path test">
  Test interruptions, silence, ambiguity, unsupported requests, prompt injection, opt-outs, wrong-party contact, provider failure, action failure, and human handoff. Review the stored activity or recording as well as the live experience.
</Callout>

Continue with [agent configuration](/chat-and-voice/agents), [knowledge and chat widget](/chat-and-voice/knowledge-and-chat-widget), and [phone, outreach, dialer, recordings, and compliance](/chat-and-voice/phone-outreach-and-compliance).