CHALLENGE
A real operational problem, not a demo.
Dispatchers juggle priorities, technician skills, time windows, sick days, and locked appointments by hand. An optimiser can make the maths faster, but a plan nobody understands is a plan nobody trusts.
SOLUTION
A system built around the work.
For a field-service CRM, BXTrack paired a vehicle-routing solver with an LLM. The solver does the maths, the model explains the proposed changes in plain language, and the dispatcher can approve or dismiss every suggestion.
WHAT MAKES THIS DIFFERENT
The automation stays accountable.
The planner is designed around human control. Locked visits remain fixed, business rules are explicit, and the AI never silently rewrites a schedule. It earns trust by showing what changed and why.
HOW IT WORKS
A clear path from signal to action.
01
Classify
Visits and staff are filtered by status, priority, skills, and availability.
02
Optimise
OR-Tools solves the multi-depot routing problem within those constraints.
03
Explain
Claude writes why each suggested change improves the plan.
04
Approve
The dispatcher approves or dismisses the proposal; the calendar updates live.
RESULTS
From fragmented work to a plan people can use.
The proof of concept runs against the client's production schema to test the approach with real operating constraints. Dispatchers receive a plan that is both mathematically sound and explainable enough to act on.