Copilot, not autopilot
Trying to fully replace the planner from day one explodes scope and exceptions. A copilot generates a proposal and the planner adjusts it: lock operations that shouldn't move, change priorities, add a rush order, simulate a breakdown, compare two scenarios and approve the plan before it reaches the floor.
Why it lowers rollout risk
The copilot captures knowledge that isn't in the ERP yet (that rule only the shop lead keeps in their head) and keeps the person in control. Autonomy can come later, once trust and data allow it.
What it does and doesn't
- Does: propose sequence and machine assignment, show orders at risk and the cost of each alternative.
- Does: reschedule on rush orders, breakdowns, missing material or delays.
- Doesn't: replace the ERP or control PLCs.
- Doesn't: make decisions without your approval.
FAQ
How is it different from a classic APS?
An APS aims to automate all planning and is usually expensive and slow to roll out. The copilot focuses on the shop's real problem — finite capacity and rescheduling — with the person at the center.
Is the edge in the algorithm?
Only partly. The engine (e.g. with OR-Tools) is powerful, but the real work is turning imperfect ERP data into a usable model and encoding each factory's rules.