> ## Documentation Index
> Fetch the complete documentation index at: https://maxim-gtm.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# How can I Integrate Maxim to Trace and Evaluate My Voice Agents?

> Maxim provides native integrations for popular voice agent frameworks, enabling comprehensive tracing and evaluation of your voice AI applications through simple one-line integrations.

## LiveKit Integration

For voice agents built with LiveKit, Maxim offers a streamlined integration:

* **One-line setup**: Instrument your entire LiveKit agent with `instrument_livekit(logger, on_event)`
* **Automatic tracing**: Captures agent conversations, function tool calls, and performance metrics in real-time
* **Complete transcripts**: Records system instructions, user messages, and assistant responses
* **Framework support**: Works seamlessly with both OpenAI Realtime and Gemini models within LiveKit

## OpenAI Realtime API Integration

Maxim supports direct integration with OpenAI's Realtime API:

* **Minimal code changes**: Use the `MaximOpenAIClient` wrapper to enable comprehensive observability
* **Session enrichment**: Add custom session names, tags, and IDs through additional headers
* **Complete traces**: Captures multi-turn conversations, tool calls with arguments and results, plus detailed timing and cost metrics

## Continuous Evaluation

Once integrated, run continuous evaluations on your voice agent recordings and transcripts using specialized voice evaluators:

* Abrupt disconnection detection
* Signal-to-noise ratio measurement
* Sentiment analysis
* User satisfaction tracking
* Word error rate (WER) calculation

Combined with Maxim's observability suite, you can identify conversation patterns, debug performance issues, and continuously improve your voice agents based on real interaction data.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.