Platform AI Platform AI
LOGISTICS

Logistics operator

Routing, telemetry and document flow for waste collection — routes planned by algorithm instead of by hand, with the result measured in vehicles taken off the road.

SECTOR

Logistics and waste collection

SCOPE

Route optimisation · telemetry · machine vision · electronic documents

ROLLOUT

Regional waste operators

ENGINE

In-house routing algorithms and telemetry server

Logistics operator

TASK

Plan and optimise collection routes across a network that changes daily — and rebuild them on the fly when it does, including complex multimodal routes.

Replace the separate tracking, document and analytics tools operators were running with one system, sitting on telemetry of our own rather than a third-party service.

RESULT

250 → 125vehicles needed after rollout at one regional operator — half the fleet
400 → 190vehicles at another operator: a 52.5% cut
2–4people now run the logistics department
+17%faster route calculation than the routing engines it replaced
3–5%more accurate routes than the same engines
500K/stelemetry requests handled per second

What we built

01

Route planning and optimisation that returns an answer in fractions of a second, however complex the route, and rebuilds it live when the plan changes.

02

A telemetry server of our own, with fast integration to satellite trackers — replacing the separate vehicle-tracking systems operators used to run alongside everything else.

03

A telemetry store and processing module handling up to 500,000 requests a second and tens of millions of rows, extendable as new tasks appear.

04

Machine vision over existing cameras and fill sensors: load and fill control for warehouses, silos, containers and truck compartments, plus image recognition over live streams or an existing photo archive.

05

Electronic document flow — built from scratch or integrated into what is already there — with e-signature, approval tracking, financial operations and automatic reporting.

06

Business analytics that surfaces the problems nobody had noticed and suggests the next step for optimisation.

Run something like this?

Tell us the load, the rules and the deadline — we will come back with a plan and a live environment.