Sessions

A trace is one interaction (a root span plus its child steps); a session is the conversation those interactions belong to - every trace sharing a session_id. Sessions catch the failure mode single-trace monitoring can’t see: every individual reply looks fine, but the conversation as a whole goes in circles, contradicts itself, or never resolves. Grouping is opt-in per call - pass any stable per-conversation id:

with client.tracer.trace("support-agent", session_id=conversation_id) as span:
    ...

Traces sent without one get their own auto-generated session.

The Sessions view

Observe → Sessions lists each conversation with:

Column	Meaning
Session	The conversation’s id
Turns	Root interactions in the conversation
Duration	Active time: the sum of the conversation’s turn latencies - not wall-clock span
Last activity	When the conversation last saw a trace
Session score	The lowest verdict among enabled session evaluators and session-scoped scorer groups that scored it, with the judge that gave it named

Failed spans anywhere in the conversation surface as an error-count badge inside the Session cell. Opening a session shows every turn in order; each turn’s full span tree is one click away.

screenshot of the session detail dialog - a list of user/agent turns in order with per-turn latency, a Judge scores section showing two evaluator verdicts with ratings, and an 'Add to dataset' button in the header. — A session's detail view: turns in order, judge verdicts, and Add to dataset.

Conversation-level judging

Three kinds of scorer judge a session as a whole, all triggered automatically when the conversation goes quiet:

Session Baseline Judge - built-in: goal progression, consistency, non-repetition. Also runs on demand from the session’s detail view. Its rubric lives in an LLM judge scorer (tunable via Tune judge); pause it from the Scorers page’s row toggle.
Session-scoped judge scorers - your own criteria with scope="session", judged on idle and re-judged if the conversation resumes. See Multi-Turn Session Evaluation.
Session-scoped scorer groups - several scorers (judges, patterns, code) blended into one 0-10 verdict per conversation, on the same idle trigger, with must-pass gates.

A verdict below the evaluator’s alert threshold raises a signal in the normal triage queue. These are also callable from the SDK - run the coherence check on demand and read the session’s spans (e.g. to walk a reported drift span up to the turn it belongs to):

score = client.monitor.sessions.coherence_check(session_id)   # one judge call
print(score["rating"], score["justification"], score["driftSpanId"])

spans = client.monitor.sessions.spans(session_id)             # roots + children, oldest first

Where sessions come from

Source	How
SDK	`session_id=` on `tracer.trace(...)`
Framework integrations	Pass `session_id` to the handler/processor constructor
OpenTelemetry	`session.id`, `gen_ai.conversation.id`, or `agentx.session_id` span attributes
Importers	`agentx-moveworks` (one session per conversation) and `agentx-databricks` (`mlflow.trace.session` metadata)
Playground	Conversation simulation records each run as a real `sim-<id>` session

From finding to regression test

A failed conversation is a future regression test: Add to dataset in the session detail turns the whole conversation into a multi-turn golden dataset case - see Datasets from Production.

Tracing

Framework Integrations

The Sessions view

Conversation-level judging

Where sessions come from

From finding to regression test

​The Sessions view

​Conversation-level judging

​Where sessions come from

​From finding to regression test

The Sessions view

Conversation-level judging

Where sessions come from

From finding to regression test