Skip to main content
Once your application is instrumented, every LLM call lands in to11 as an OpenTelemetry-native trace. Observe is where you read that data back — to follow a single request, debug a failure, or understand how a multi-step workflow actually ran. Data is scoped to a project: pick a project, and you see its traces. Today, Observe means traces. Aggregate metrics and pre-built dashboards build on the same data and are on the way.

​
Traces

A trace is the full record of a request — every LLM call with its inputs and outputs, latency, tokens, and cost, plus any child calls nested under one trace for a multi-step workflow. The Trace Explorer lets you filter by time, provider, model, status, and more, then open any trace to inspect its spans in a waterfall view.

​
Metrics and dashboards

Coming soon. Aggregate latency, token, error, and cost signals over your traces, plus pre-built dashboards for your golden signals.

​
Access

Who can view traces — and production traces in particular — is governed by your role. See Roles & permissions.

​
Next steps

Traces

Explore and filter individual requests.

Instrument

Get your data flowing first.