“Behind the green building, near the temple” is a real address. Geocoding is a probability problem, not a lookup.
Driver apps that work for a Tamil-speaking trucker and a Hindi-speaking 2W rider — same codebase, no compromises.
30–40% of inbound calls are status-checks. They cost real money on a per-call basis.
Real-world constraints (vehicle type, time windows, driver skill, return-leg load) break textbook OR-tools setups.
Cross-platform iOS + Android driver apps with offline-first sync, voice-driven UI for low-literacy users, and embedded route nav.
Pragmatic VRP solvers tuned for Indian roads — turn restrictions, road conditions, vehicle constraints, return-leg matching.
Multilingual voice AI for status-check calls, address verification, COD escalations. Hybrid AI + human, billed per resolved call.
Agentic copilots for dispatch operators — surface delays, suggest reassignments, draft customer comms.
Cross-platform Flutter app — offline-first, voice-driven, embedded routing. Replaced 3 legacy apps, 12K DAU at peak.
VRP solver tuned for hub-and-spoke + last-mile constraints. 28% km saved without dropping SLA hits.
Yes — that's the design constraint. Offline-first sync, sub-50MB install, voice-driven where literacy is a constraint. We test on Android Go devices.
Yes. We expose a route-engine API and integrate with FarEye, Locus, Shipsy, or your custom TMS via webhooks.
Per-resolved-call. Volume-tiered. Status-checks are typically the cheapest unit; COD escalations and address-verification cost more.