01 · Capture

Staff opens the capture sheet.

BYOD smartphone. One button. No PII typed. Speak (or paste) and the incident is captured, classified, and routed — staff makes no in-the-moment decision.

17:42● ● ● ● 92%
事案を記録
VOICE · READY
Privacy: capture runs on-device. Raw audio retained 30 days. Stop any time.
Voice capture needs Chrome/Edge. Type or paste meanwhile.
What happens technically
  • Speech transcribes to text on the device (here: browser Web Speech API)
  • Rule pre-filter triggers instantly on obvious threat keywords
  • Transcript posts to /api/incidents → classified + persisted
  • Severity routes the incident to the right escalation tier
  • Staff sees "logged, manager notified" — no in-moment decision needed
What does NOT happen
  • Customer is not notified — capture is for the record, not confrontation
  • Customer face / image not stored
  • Staff is not asked to classify on the spot
  • No PII (name, card number) extracted or persisted
Pressing Capture & classify creates a real incident and takes you straight to its evidence packet + escalation controls.
Prefer the classifier sandbox? →Back to demo index