Last-mile chaos tamed
A 3PL was burning fuel and missing ETAs. They needed routes that learned and delays they could see coming.
A 3PL was burning fuel and missing ETAs. They needed routes that learned and delays they could see coming.
Dispatchers built routes by experience and spreadsheets. Traffic and weather showed up as surprises. Customers complained about late or vague ETAs. The 3PL had telemetry and order data but no single system that could optimize the next day’s runs and update ETAs as conditions changed. They needed to cut cost per drop and improve reliability without replacing their TMS overnight.
We built an optimization and prediction layer that ingests orders, capacity, and constraints from their TMS, plus traffic and weather feeds. It produces daily route plans that minimize distance and time while respecting time windows and vehicle limits. A separate delay model takes real-time progress and external signals and updates ETAs so customers and ops see a single, improving estimate. Drivers get turn-by-turn and dynamic adjustments when something changes. We integrated via API so their existing TMS and driver app remained the source of record.
Miles per delivery dropped; fuel and overtime costs followed. ETA accuracy improved and “where’s my order?” calls fell. The 3PL is now using the same engine to test different fleet sizes and service areas before committing to new contracts.