Technical overview
One stack. Four execution surfaces.
CityTriage AI is not a point solution—it is a connected architecture. Citizens capture reality in the field; councils orchestrate policy and geography; operators optimise assets across depots; drivers close the loop with proof. The sections below trace how data and decisions flow across B2C, B2G, B2B, and B2D without brittle integrations.
Layer 01
B2C Citizen Portal
The citizen surface is deliberately lightweight: a frictionless web experience that works on any device, with no app-store funnel for time-critical reports. Uploads are normalised server-side so Vision AI can classify hazards, estimate scale, and attach structured metadata before a human ever opens a queue.
That ingestion contract—image plus context, returned as machine-readable triage primitives—is what lets downstream B2G logic stay deterministic. Citizens get clarity; the platform gets signal instead of noise.
- Progressive web patterns for sub-second first paint on mobile networks
- Vision pipelines for categorisation, sizing hints, and confidence scoring
- Immutable event envelope handed to the command tier for auditability
Layer 02
B2G Command Center
Councils operate under policy, statute, and political risk simultaneously. The B2G layer encodes your rules as automated logic gates: severity thresholds, ward ownership, escalation paths, and SLA timers execute consistently on every inbound event—so supervisors review exceptions, not every line item.
Mapbox-backed triage surfaces put those decisions on a live map: hotspots, fleet proximity, and historical recurrence inform prioritisation in one view. Geography is not decoration; it is the coordinate system the rest of the stack inherits for routing and accountability.
- Configurable gate chains (validation → risk score → assignment)
- Map-first operations with layers for assets, crews, and open work
- Audit trails suitable for democratic services and FOI-ready exports
Layer 03
B2B Commercial Logistics
Waste hauliers and logistics partners rarely run from a single yard. The B2B tier models multi-depot reality: capacity, shift patterns, vehicle class, and contractual SLAs are first-class inputs. Telematics streams—position, idle time, and utilisation telemetry—continuously reconcile what the plan promised with what the fleet is doing.
When B2G promotes work into execution, B2B decides which depot, which run, and which asset class absorbs the job without breaking round continuity. The same graph powers scenario planning when you rebalance routes after a vehicle drops out mid-shift.
- Depot-aware rostering and cross-dock style hand-offs where required
- Live telematics ingestion for deviation detection and ETA refresh
- Commercial KPI roll-ups: tonne-km, cost-to-serve, and carbon proxies
Layer 04
B2D Driver Execution
Drivers are the final mile where plans meet physics. The B2D experience is built for glare, gloves, and interruption: turn-by-turn aware of live re-optimisation when a new high-priority job enters the polygon. Dynamic rerouting is not a map gimmick—it is the feedback loop that protects SLAs when the world changes between depots.
Digital proof of delivery (PoD)—timestamp, geo-stamp, and media where appropriate—closes the chain of custody back to B2G and B2B. Disputes shrink; contractor billing aligns with verified completion; residents see progress they can trust.
- Driver-native UI with offline-tolerant task state where networks fail
- In-shift reroute payloads minimising unnecessary U-turns and dead mileage
- Structured PoD events that reconcile automatically to work orders