Resource · Encounter reports

What do encounter reports look like?

Every encounter a client has with our behavioral health AI agents produces a structured report. Appointment check-ins, between-session check-ins, intervention calls, and companion conversations each become a measurable, longitudinal view of the client between visits, ready for the clinician. Here's what's inside.

Voice biomarkers

Check-in reports

Comparative progress

Predictive risk analysis

Physiological indicators

Voice biomarkers

A behavioral health screen at every AI agent encounter.

Our voice biomarkers look for specific speech patterns and signals to understand client behavior during every AI agent encounter. One of the patterns we score is depression and anxiety severity.

Built on top of Kintsugi Health's open-source voice biomarker model (hosted on Hugging Face), we calculate these scores to give behavioral health clinicians a measurable check-in data point at every encounter, whether it's an appointment check-in agent, an intervention agent call, or a quick exchange with our companion agent.

Meet your agents

Voice biomarker readout

Appointment check-in

Depression

0.34

/ 1.00

None

Mild

Mod

Severe

Anxiety

0.62

/ 1.00

None

Mild

Mod

Severe

Scores derived from speech signal. Not a diagnostic instrument; informs the clinician's review.

Check-in reports

The week between sessions, in one behavioral health report.

Every between-session check-in with our companion AI agent captures challenges, wins, and important moments, creating a detailed behavioral health report the clinician can read before the next session starts.

Instead of asking "so, how was your week?" and rebuilding context for ten minutes, clinicians walk into the next session already knowing what happened, what worked, and where the client got stuck.

Important moments

For your session on Friday, May 17

Win

Mon, 9:42 AM

Made it through the team presentation without spiraling. Used the breathing exercise from last week before walking into the room.

Challenge

Wed, 11:15 PM

Hard night. Couldn't sleep, kept replaying the conversation with my brother. Felt like I was back in old patterns.

Moment

Thu, 6:30 PM

Noticed I was getting overwhelmed at the grocery store and stepped outside for 5 minutes. First time I've caught it that early.

Surfaced automatically from 6 check-ins this week.

Comparative progress

Behavioral health, finally with data-driven progress metrics.

Behavioral health has long lacked data-driven metrics. Every Vybz AI agent encounter shows how things have changed across the life areas that matter, compared to the past.

Work, relationships, family, sleep, and the other domains that shape a client's day surface as quantifiable progress signals in every behavioral health report. Clinicians get insights that unlock the important areas of the client's life, not just a symptom score.

Comparative progress

Life areas, 30 days

Work

0.71

+0.39

Relationships

0.68

+0.27

Family

0.49

-0.06

Sleep

0.62

+0.24

Synthesised from 18 encounters across the last 30 days.

Predictive risk analysis

Patterns that show what's coming, not just what just happened.

Every Vybz behavioral health AI agent encounter feeds a predictive risk model that watches for the patterns clinicians already know matter, before they become a crisis.

A sudden drop in activity, social withdrawal, sleep disruption, or a drift in voice biomarker scores can each highlight the risk of a depressive episode. We surface the early signal, name the signals that drove the forecast, and route it to the clinician so the next intervention is timed, not reactive.

Risk forecast

Next 14 days

Depressive episode

Confidence 0.74

Elevated

Watch

Driving signals

Activity decline

Daily movement down 38% over 10 days.

Social withdrawal

Initiated contact 1× this week, vs. 6× baseline.

Sleep disruption

Onset latency up to 52 min, three nights running.

Voice biomarker drift

Depression score trending up across last 4 encounters.

Forecast is informational. The clinician decides the next step.

Physiological indicators

The body tells the story the client can't always put into words.

Our behavioral health AI agents plug directly into Google Health, Oura, Fitbit, and other wearables to access the physiological metrics clients are already tracking, with their consent.

Resting heart rate, heart rate variability, sleep stages, daily steps, and other wearable signals give every encounter a physiological context that self-report alone can miss. The same metrics feed the predictive risk model, so behavioral and physiological shifts can surface a depressive episode, anxiety spike, or relapse risk earlier than either signal could on its own.

Physiological indicators

7-day window

Resting heart rate

bpm

via Google Health

Heart rate variability

−11

via Oura

Sleep

6.2

hrs

−0.8

via Oura

Daily steps

3,810

/ day

−38%

via Fitbit

Read-only access. The client controls what their wearable shares.

How does the report help?

Better care starts with better signal between sessions.

Context-rich out-of-session data

Encounter reports capture what's happening between visits, so the next session and the broader care plan are built on signal, not on what the client can recall in the room.

Captures moments missed in session

Wins, challenges, and turning points surface as they happen, not reconstructed days later. The clinician walks in knowing what mattered this week.

Surfaces early warning signs

Behavioral and physiological shifts come together in one feed, so a developing depressive episode, anxiety spike, or relapse risk reaches the clinician before it becomes a crisis.

See it for yourself

Book a demo to see how the report helps.

Walk through a real Vybz encounter report with our team. See voice biomarkers, check-ins, comparative progress, risk forecasts, and wearable signals in one clinician-facing view.

Book a demo