kanari

Your voicecan hintbefore you notice.

kanari estimates stress + fatigue signals from your voice patterns, tracks them over time, and turns rising risk into small recovery actions you can schedule. Not medical advice.

Start check-in
Scroll

30s

daily voice check-in

On-device

acoustic features

Evidence

quotes + breakdown

3-7

day trend forecast

The Problem

You don't notice burnout until it's too late.

Burnout creeps up silently. By the time you feel exhausted, overwhelmed, or detached—the damage is already done. Self-assessment fails because burnout impairs the very self-awareness needed to detect it.

Your voice tells a different story. Speech patterns, pause frequency, vocal energy—these biomarkers can shift before you consciously feel the strain.

66%

report experiencing burnout (U.S., 2025)[1]

Burnout often shows up late because self-assessment fails when you are already depleted.

Voice

can reflect stress-linked acoustic shifts[2]

Research links stress with changes in pitch, pauses, and energy variability. Kanari treats these as signals, not a diagnosis.

Kanari turns a 30-60s voice check-in into trends and small recovery actions you can schedule. Not medical advice.

Burnout is common. Most people only recognize it after the crash.

kanari listens. Estimates. Acts.

Product

From a 30-second voice check-in to a recovery plan.

Gemini Live runs the conversation. On-device biomarkers track trends. Gemini Flash synthesizes evidence-linked insights and lightweight actions you can schedule.

Privacy-first. Browser-based. Powered by Gemini 3.Gemini

Voice Biomarker Analysis

Estimate stress and fatigue signals from speech rhythm, pauses, and vocal energy. Acoustic feature extraction runs in your browser.

Predictive Forecasting

Compute a short-horizon (3–7 day) risk forecast from your recent trend. Heuristic signal, not a diagnosis.

Longitudinal Tracking

Build your personal baseline over time. Understand your patterns and catch deviations before they become problems.

Calendar Integration

Schedule recovery blocks when risk is elevated. Small interventions beat a full crash.

Walkthrough

See it in action.

A quick, visual tour of the full flow: check-in, on-device biomarkers, insights, 3-7 day forecast, recovery actions.

1/7
Loading...

Feature

Voice check-in (Gemini Live)

A 30-60s conversation that feels natural, not like a form.

What you see

  • Real-time streaming with Gemini Live (mic -> conversation -> live responses)
  • Captures session + transcript for later review in History
  • Designed for daily use: low friction, fast feedback

Trust

No black box.

Kanari is built to show its work: what is computed on-device, what is sent to Gemini, and why a session does (or doesn’t) count toward your trend.

On-device featuresEvidence-linked quotesQuality gated trends

Tip: hover or tap the underlined words for quick definitions.

Trust drawer

Expand any item. This matches what you can inspect inside a saved check-in.

See it in app

Kanari extracts in your browser and stores sessions in .

Local-only

  • Feature extraction (rhythm, pauses, energy, spectral features)
  • Baseline + calibration math
  • Trend aggregation + 3–7 day heuristic forecast

What you can inspect

  • Per-feature values + thresholds/weights
  • Intermediate scores (acoustic, semantic, blended)
  • Confidence + data quality

Not medical advice. Kanari estimates patterns and trends; it does not diagnose burnout.

Prevent the crash.

Your voice carries signals you may miss. Let it help you plan recovery.

Try kanari
A glowing crystal with orbital lines in a smoky atmosphere

Built for Google DeepMind Gemini 3 Hackathon 2025