> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.rubrichq.io/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.rubrichq.io/_mcp/server.

# Integrations Overview

RubricHQ tests your conversational agent by running an **AI-simulated caller** against it. You define a **scenario** (the caller's persona, goal, and step-by-step instructions) and a set of **metrics**, then RubricHQ runs the conversation and evaluates the transcript.

The only thing that changes between integrations is **how RubricHQ reaches your agent** — the *transport*. Everything else (scenarios, personas, metrics, results) is shared.

## Choose a channel

A RubricHQ agent has up to three channels. You pick which one a test run uses.

| Channel         | When to use                           | Integrations                   |
| --------------- | ------------------------------------- | ------------------------------ |
| **Phone**       | Your agent answers a phone number     | Twilio                         |
| **Web (voice)** | Your agent streams audio over the web | Vapi, Retell, LiveKit / Custom |
| **Text**        | Your agent exchanges text messages    | Custom WebSocket (X-RUBRIC)    |

> **Note**
>
> **Text testing is \~10× faster and far cheaper than voice.** Use it for workflow validation, regression testing, and CI. Reserve voice testing for ASR/TTS validation and final pre-prod checks.

## Voice vs Text — what's evaluated

* **Voice** runs produce an audio recording, so you get **transcript metrics** *and* **audio metrics** (latency, interruptions, words-per-minute, voice clarity, etc.).
* **Text** runs have no audio, so only **transcript metrics** apply. Audio-only metrics are automatically skipped for text runs.

## Quickstart

1. **Create an agent** and open **Channels**.
2. Configure the channel you want to test (Phone, Web, or Text) — see the per-integration pages.
3. **Create a scenario** for that agent (persona + instructions + outcome) and attach metrics.
4. **Run** it — from the Evaluator / Batch Run page, choose the channel and click **Run**.
5. Review the **transcript, metrics, and (for voice) recording** in the results.

Next:

* **[Text Testing](/integrations/text-testing)** — connect your agent over the Custom WebSocket (X-RUBRIC) protocol, with a runnable example.
* **[Voice Testing](/integrations/voice-testing)** — overview of every voice integration (phone + web).
* **[LiveKit](/integrations/livekit)** — connect your own bot in a LiveKit room (client- vs RubricHQ-managed rooms), with request/response examples and room-config keys.
* **[Daily](/integrations/daily)** — connect your own bot in a Daily room; same two modes, with Daily room-properties (incl. recording) keys.
* **[Vapi](/integrations/vapi)** — test an existing Vapi assistant; sync or paste instructions, auto-fetch calls.
* **[Retell](/integrations/retell)** — test an existing Retell agent; sync or paste instructions, auto-fetch calls.