Transport route optimization
We decide which vehicle carries which orders, in what sequence and along which roads. Fewer units on the street, fewer kilometers per delivery and fewer hours of running engine — with artificial intelligence that explains every decision.
The starting point
Urban delivery concentrates the worst conditions for fuel consumption: heavy vehicles, constant stop-and-go, traffic, and routes that are usually decided by habit rather than by calculation. A truck that leaves half loaded, a visit sequence with crossings and backtracking, or a delivery that has to be repeated the next day: all of it is paid for twice, in diesel and in emissions.
Our service attacks that exact point — not with general recommendations, but by solving the mathematical problem of assignment and sequence over the real road network, before the first truck starts its engine.
What we use
We do not estimate straight-line distances: we work over the real road network, with traffic directions, turn restrictions and traffic conditions.
Decides how many vehicles go out
It solves the capacitated vehicle routing problem: it receives the distribution center, the geolocated orders with weight and volume, and the available fleet with the real capacity of each unit. It returns which vehicles go out, with what load and in what sequence — also respecting the maximum number of stops per shift.
Sequences the route
It takes an already assigned delivery run and reorders its destinations by the sequence with the lowest real driving distance and time. This is the operation that removes the crossings and backtracking typical of a manually built route, and returns the exact path to draw on the map.
Enables comparison before operating
It calculates point-to-point distances and times between every origin and destination. This is what allows several delivery scenarios to be simulated and the best one chosen before committing the shift — without moving a single truck to find out.
How we combine it with AI
A route nobody understands is a route the supervisor ends up changing by hand — and that is where the savings are lost. That is why every result can come with an explanation in plain language, generated by AI over the exact data the optimizer produced.
“This assignment is efficient because a single vehicle was used to cover all the orders, reducing costs and resources. In addition, the stops are grouped geographically in an optimized route, minimizing the total distance of 38.9 km and the total time of 96 minutes.”
The decision belongs to the optimizer; the AI only narrates it. No truck moves because of what a language model says.
It receives the kilometers, the minutes and the vehicles already calculated, and it is explicitly instructed not to add data that is not in the input.
If the explanation fails, the route is delivered anyway. It is added value, not a critical dependency.
The environmental impact
The environmental benefit of optimizing routes is not a side effect of the operational savings: it is exactly the same fact, measured in liters instead of dollars.
Highest impact
By consolidating into a single unit the load that used to be split across several, there are trucks that simply do not go out. A vehicle that does not go out does not save a percentage of fuel: it saves one hundred percent of its own, plus its wear and its use of public roads that day.
Direct impact
The optimal sequence removes the crossings, backtracking and zigzags of the manually built route. Every kilometer not driven is fuel not burned, and every liter of diesel not burned is 2.68 kg of CO₂ that is not emitted.
Sustained impact
The engines optimize driving time, not only distance. Fewer hours in congestion mean less idling — the state in which a vehicle consumes fuel without moving a single meter.
Cumulative impact
The routes generated are feasible: they respect real capacity and maximum stops, and orders that do not fit are reported instead of forced in. Fewer failed deliveries means fewer re-deliveries, and a re-delivery doubles the footprint of that order.
How the savings are calculated
We publish the method before the result. Savings are calculated over the kilometers returned by the optimizer itself and the fuel-consumption parameters of each fleet — not over industry averages.
km avoided = (km manual route − km optimized route) + (vehicles avoided × average route km) liters avoided = km avoided × fleet consumption (L/100 km) / 100 kg CO₂ avoided = liters avoided × fuel factor
| Parameter | Value |
|---|---|
| Diesel — direct combustion | 2.68 kg CO₂/L |
| Gasoline — direct combustion | 2.31 kg CO₂/L |
| Medium delivery truck | 25–35 L/100 km |
| Delivery van | 10–14 L/100 km |
About the figures. The emission factors come from the GHG Protocol and the DEFRA inventory (scope 1). Fleet consumption is calibrated with the client's data, and the kilometers come from each run of the optimizer. We prefer to deliver a number that each client can audit rather than a generic brochure figure.
How it is built
Every optimization is recorded with its volume and its real cost, by client and by period.
Provider keys live encrypted in a single place, never scattered across each installation.
Each installation has its plan, its permissions and its quotas. It is enabled or cut off without touching the application.
The contract is ours, not the map provider's. Switching engines is configuration, not a new development project.
In short
Optimizing delivery is the most profitable sustainability decision a logistics operation can make: it cuts cost and footprint with the same move, without changing the fleet or the infrastructure. You just have to stop deciding routes out of habit.
Let's talk about your operation