Session Insights

Every Support Call, Interview, or Test — as a Structured Emotion Report

Record any opt-in interaction — a support call, a user interview, a usability test — and get back a structured report of how the person actually felt: where frustration spiked, where trust built, and how the session ended. Built on the same detect · update · adapt models as our real-time products.

See a sample report

Any modality

voice · video · text

Structured report

not a raw transcript

Opt-in only

consent-gated by design

Proof Block

The output is a report, not a raw signal dump

Every session ends with a structured summary your team can act on — not a transcript someone still has to read end to end.

Live sample

Every call in the Voice Agent demo ends with one

Run the live demo on this site — the session report shown at the end is generated the same way.

Emotional arc

opening vs. closing state, plus the dominant signal

Not a single sentiment score — a timeline of how the state moved across the session.

Key moments

adaptation points flagged automatically

Turns where the detected state crossed a threshold are surfaced without manual tagging.

Example output

Sample session report

{
  "session_id": "call_9f21",
  "duration_s": 214,
  "dominant_emotion": "anger",
  "distribution": {
    "anger": 0.38,
    "neutral": 0.31,
    "happy": 0.19,
    "fear": 0.12
  },
  "opened": "neutral",
  "closed": "happy",
  "arc": "improving",
  "adaptations": 3
}

JSON

From recording to something a team can use

Three steps: capture, analyze, deliver.

Capture

Any Opt-In Session

Point it at a support call, a moderated usability test, or a recorded interview — voice, video, or text transcripts all feed the same pipeline.

›Voice, video, and text input
›Works alongside existing call/recording tools
›Consent and redaction built into the intake step

Analyze

Turns Raw Signal Into a Report

The same face, voice, and text models — and the fusion engine — behind our real-time products run once, after the fact, across the whole session.

›Emotional arc across the entire session
›Key moments flagged automatically
›Dominant emotion with a confidence score, not just polarity

Deliver

A Report Your Team Can Actually Use

Structured output built for QA leads, researchers, and CX teams — not a data scientist's notebook.

›Per-session report export
›Aggregate trends across many sessions
›Plugs into existing coaching or research workflows

Where it fits

🎧

Support QA & coaching

See which calls actually needed a supervisor, and why.

🧑‍🔬

User research

An emotional read on every usability test, not just what testers said out loud.

🗣️

Customer interviews

Know which parts of the pitch actually landed.

🏢

Enterprise CX programs

Trend emotion across thousands of interactions, not just a CSAT score.

See what a session report actually looks like

Try the live voice agent demo end to end, then talk to us about wiring this into your own call or research pipeline.

Talk to David