Airlines lose most to irregular operations, a delay, a tech stop, a weather closure - when the schedule must be rebuilt under pressure, fast. Our engine re-optimises routing and fleet assignment in milliseconds, and matches the provably optimal answer where optimality is provable. It is the same mathematics that beats Google OR-Tools on the public VRP benchmarks we ran, applied to the operations that move metal and people.
Z3-certified optimal where optimality is provable (5/5 on exact instances); 3.05% mean from the world's best-known on the public routing benchmark, beating Google OR-Tools on every instance - with the margin widening on the hardest problems. Every result is certificate-verifiable; the method is proprietary. Full report: TR-2026-01.
What runs live on the engine today, and what is a scoped enterprise build.
Sequence flights to tails respecting minimum turnaround, curfews and station times, as a time-windowed routing problem the engine solves to the best run in seconds and re-solves in milliseconds.
Catering trucks, fuel bowsers, baggage tugs, de-icing rigs and crew buses across a hub, a multi-vehicle, time-windowed, capacitated schedule: exactly what the API solves and certifies today.
When the day breaks, re-optimise the affected routing and assignments from the live state in milliseconds, the engine's defining strength is that re-planning costs the same as planning.
Moving crews between hotels, terminals and aircraft on time, at lowest cost, a routing problem with hard time windows, solved and re-solved live.
Assignment-class problems we scope and model on the same energy-optimisation core, validated against your historical schedules before any go-live.
A distinct set-partitioning class. We engage consultatively, model it on the core, and prove the result against your incumbent before deployment, never claimed as solved before it is measured.
| The operational pain | What the engine delivers |
|---|---|
| IROPS recovery is slow, controllers rebuild by hand while costs compound | Millisecond re-optimisation from the live state; options on screen before the next decision |
| Ground-fleet schedules are built once and degrade all shift | Continuous re-optimisation, affordable to re-run on every gate change or delay |
| Optimisation vendors are black boxes you cannot audit | Every solution is a verifiable certificate; quality is benchmarked publicly, not asserted |
| Data-sovereignty and security constraints block cloud tools | On-premise / private-cloud deployment; your operational data never leaves your tenancy |