The Challenge
The operator moved freight across many corridors but had no unified view of where shipments were. Routing was decided by habit, not analysis, so multi-leg journeys were often more expensive and slower than necessary. When a leg was delayed, no one knew until a customer complained.
The Solution
We built a logistics control tower on the Baalvion Operating System. Every shipment milestone publishes an event, giving end-to-end live visibility. A multi-leg route optimiser models the corridor network as a lane graph and searches it for the cheapest, fastest, or balanced route — with fallback rules that relax constraints rather than failing when an ideal route doesn't exist. It builds on the same event backbone as our automated customs clearance.
- Live tracking from event-driven shipment milestones.
- Lane graph modelling carriers, legs, and corridors.
- Route optimiser producing cheapest / fastest / balanced options.
- Fallback rules that relax constraints instead of failing.
Architecture
Shipment milestones flow in as events and update a live tracking view. The optimiser represents the network as a lane graph — nodes are hubs, edges carry carrier rates and transit times — and runs a bounded search to produce ranked route options. When constraints can't all be satisfied, fallback rules synthesise a relaxed route rather than returning nothing. This is the multi-leg counterpart to single-leg quoting, and feeds the logistics sector solution.
Technology Stack
Event-driven milestone tracking, a lane graph, a bounded route optimiser, pluggable carrier-rate providers, and a real-time dashboard — built through our automation and cloud solutions practices.
Results
The operator gained end-to-end shipment visibility for the first time. Optimised routing cut route cost by roughly 18%, and on-time delivery improved by 12 percentage points as delays were spotted and rerouted around in real time.
Lessons Learned
Modelling the network as a graph made optimisation tractable and explainable. Fallback rules mattered as much as the optimiser — in the real world an ideal route often doesn't exist, and returning a good-enough one beats returning nothing. And visibility had to come first: you can't optimise what you can't see.