Build a Custom Route Optimization Algorithm for Your Fleet

Many local delivery businesses rely on the basic routing functions inside their TMS, like Samsara or Motive, or even manual planning in Google Maps. These tools are designed to find the shortest path between two points or for a single vehicle's stops. They cannot solve the complex Vehicle Routing Problem (VRP) which involves multiple vehicles, capacities, delivery time windows, and driver hours.

Consider a local furniture company with a 15-truck fleet. A dispatcher spends the first two hours every morning manually grouping hundreds of stops into routes using a spreadsheet and a map. An urgent delivery for a key commercial client comes in at 10 AM. The dispatcher has no systematic way to know which driver has the vehicle capacity and is geographically best positioned to handle it without disrupting their entire route. This leads to 20 minutes of phone calls and a suboptimal assignment that adds 45 minutes of drive time and costs over $100 in fuel and overtime.

Off-the-shelf route planners like OptimoRoute or Route4Me offer more features but impose a generic model on your business. Their algorithms optimize for distance and time, but they cannot incorporate your specific business costs. For example, they cannot model the high cost of a late delivery to a key B2B client versus a flexible residential drop-off. The structural problem is that these SaaS tools are built for the 80% use case with a fixed data model. You cannot add a custom constraint like “this stop requires a truck with a liftgate,” forcing dispatchers back to manual overrides and negating the tool's value.

The first step is a discovery and data audit. Syntora would connect to your TMS to pull 6 months of historical delivery data to understand your specific operational patterns. We would work with your dispatcher to codify every business rule: vehicle capacities (weight and volume), driver shift times, specific customer delivery windows, average service time per stop, and any vehicle-specific constraints. This audit produces a precise specification for the optimization model before any code is written.

The technical approach uses a Python service built around Google’s OR-Tools, a production-grade library for combinatorial optimization. This service is wrapped in a FastAPI application for integration. The system would pull daily orders from your TMS API, construct a distance matrix using a cost-effective engine like OSRM, and then feed the orders and constraints into the OR-Tools solver. This approach provides full control to model the unique financial and operational realities of your fleet.

The delivered system is a background process, not another dashboard for your team to learn. The service runs on a schedule (e.g., 5 AM daily) on a serverless platform like AWS Lambda, keeping hosting costs under $50/month for most fleets. The optimized routes are pushed directly back into your TMS, appearing on drivers' mobile devices as if they were planned manually. An API endpoint allows for on-demand re-optimization of a subset of routes during the day.