Antenor Challenge
Last-mile delivery is where logistics margins go to die. Every courier shift is an instance of the traveling salesman problem: dozens of stops, time windows, package constraints and traffic, and the optimal answer is NP-hard to compute. At marketplace volume the problem compounds. Orders keep landing while routes are being planned, couriers cannot wait minutes for a solver to converge, and a route that is 15 percent longer than it needs to be is a cost paid on every single shift, in every city, every day.
Moova needed routing that was both good enough to compete on cost and fast enough to feel instant inside its courier operation.
Solution
Antenor built Moova’s route optimization engine as a low-latency API. The solver combines operations research techniques for the TSP core with machine learning heuristics that prune the search space, so a courier’s full route comes back in seconds rather than minutes. The engine is wired directly into Moova’s dispatch flow: shipments arrive, routes are solved, couriers move.
The system runs in production across seven countries: Argentina, Chile, Colombia, Guatemala, Mexico, Peru and Uruguay. Moova describes its platform as AI-powered route optimization, and that engine is this one.
Stack
- TSP and vehicle routing solvers built on operations research methods
- Machine learning heuristics to guide and prune the search
- Low-latency routing API integrated into Moova’s dispatch pipeline
- Production deployment across 7 country operations
Results
- Millions of shipments routed through the engine
- Full courier routes solved in seconds, keeping dispatch real time
- Live in 7 countries: Argentina, Chile, Colombia, Guatemala, Mexico, Peru and Uruguay
- Moova powers deliveries for Mercado Pago, Nestlé, AT&T, Samsung, Puma and UPS
- Independently verifiable at moova.io
