Skip to content
[RETURN TO ARCHIVE]CASE 04 // 04
CASE / MULTIMODAL AI & CIVIC SYSTEMS2026

Urban Incident Response.

Turns fragmented citizen reports (photographs, written descriptions, voice notes and GPS) into structured, de-duplicated, prioritised incidents routed to the department that owns them. The human-in-the-loop rules are enforced in code rather than documented: closure refuses without post-action evidence, a priority override is stored beside the recommendation instead of replacing it, and a low-confidence category is withheld into a triage queue. Every intake emits a trace of what each pipeline stage did and how long it took.

BUILT WITH

  • Next.js
  • TypeScript
  • Google Gemini API
  • Leaflet
  • Geo-temporal Dedup

LINKS