LogixFlow

Operations leaders had no single source of truth for fleet status. Routes were planned in spreadsheets and adjusted by phone — every delay cascaded.

We shadowed dispatchers for two weeks, instrumented existing tools, and identified that 80% of delays stemmed from just three reroute decisions per day.

Built around a real-time event spine: every truck, stop and exception flows into one stream that the UI and the AI optimizer both consume.

Map-first dashboard with a command-bar for dispatchers, contextual side panels, and zero-modal workflows so attention never leaves the fleet view.

Next.js on the edge for low-latency UI, Firebase for realtime sync, Mapbox for routing visuals, and a queue-driven optimizer that runs continuously.

Trained a route-scoring model on historical telemetry to predict ETAs and proactively suggest reroutes when a driver is likely to miss a window.