Observability Overview

Zero Runtime’s observability tools give you deep insight into your agent’s performance and behavior. Every session emits telemetry automatically. The Dashboard collects and visualizes it by default, with no extra setup.

What gets captured

Every agent session emits three signals:

Signal	Captures
Metrics	Latency for each pipeline stage, token usage, and cost.
Logs	The agent’s log records, filtered by `log_level`.
Traces	Spans for STT, end-of-utterance detection, the LLM, tool calls, and TTS.

Recording captures the session audio and video alongside this telemetry. See Recording.

Where to read them

Where	What you get
Zero Runtime Dashboard	Every session’s metrics, logs, and traces under Sessions, with no setup.
Metrics hooks	Each stage’s metrics in your own code as every turn completes. See Metrics.
Your observability stack	Metrics, logs, and traces shipped to Datadog, SigNoz, New Relic, or any OpenTelemetry backend through an `Exporter` on the room’s `Observability`.

What’s Next

Metrics

Read per-stage latency and token usage.

Logs

Set the log level and ship agent logs to your stack.

Traces

Follow each turn span by span and export the spans.

Recording

Capture session audio and video alongside this telemetry.

References

Python
Node JS

Examples

Observability Hooks

Collect metrics and traces from a session.

Get started

Quickstarts

Build

Testing and Eval

Observability

Deploy

Plugins

What gets captured

Where to read them

What’s Next

Metrics

Logs

Traces

Recording

References

Examples

Observability Hooks

Examples

Pipeline Events

​What gets captured

​Where to read them

​What’s Next

Metrics

Logs

Traces

Recording

​References

​Examples

Observability Hooks

​Examples

Pipeline Events

What gets captured

Where to read them

What’s Next

References

Examples

Examples