Proof of concept · runs on simulated factory data

AI Adaptive Job Shop Scheduling

A production schedule that re-plans when the shop floor changes. When a machine goes down or an urgent order arrives, the scheduler measures the impact, recommends a new schedule and a planner approves it.

Discuss a Pilot How it works

Scheduling boardScreenshot coming soonassets/promonitor/scheduler-board.webp

Adaptive scheduler Gantt board showing jobs per machine

The problem

A fixed schedule is out of date within hours

Job shops run many different orders across many machines. The plan made in the morning rarely survives the shift.

A machine breaks down

Every job planned on it is now late or blocked, and nobody knows by how much.

An urgent order arrives

Fitting it in by hand means pushing other orders back without seeing the effect.

Material arrives late

Jobs waiting for it sit idle while other machines could have been used.

Cycle times drift

Actual run times differ from the plan, and the gap grows through the shift.

How it works

Detect, measure, optimize, approve

The scheduler works as a closed loop, and a person stays in control of every change that reaches the shop floor.

1
Detect the change
Machine, job and material events arrive in real time and are checked against the approved schedule.
2
Measure the impact
The scheduler lists the affected orders and operations, the estimated delay if nothing is done, and the alternative machines.
3
Re-optimize
An optimization engine builds a recommended new schedule that keeps as many orders on time as possible.
4
Planner approves
Old and new schedules are compared side by side. Nothing changes on the shop floor until a planner approves it.

Plan
Schedule
Execute
Real-time state
Detect
Predict
Optimize
Reschedule
Learn

Capabilities

What the scheduler does

Flexible job-shop optimization

Google OR-Tools CP-SAT schedules operations across eligible machines, respecting routings, shift calendars and planned maintenance.

Configurable objectives

Weigh late orders, tardiness, makespan, setup and idle time to match what your plant values, and change the weights at runtime.

Impact analysis

Affected orders, planned versus no-action completion time and customer impact are shown the moment something goes wrong.

Old versus new schedule

KPIs, per-order completion changes and every operation move are compared before you approve.

What-if simulation

Test “machine down for 2 hours”, an urgent order or late material against the live plan without touching it.

Drag to re-plan

Drag a job to another machine or time on the Gantt board; the optimizer pins it and re-plans the rest as a recommendation.

Versions and audit trail

Every schedule version, the decision behind it and the events that led to it are kept.

Prediction models

Lightweight models estimate cycle time, breakdown risk, repair time and order completion time.

The application

Screens

Screens are from a demonstration environment that uses a built-in factory simulator with sample machines and orders.

Adaptive scheduler executive dashboard — **Executive dashboard**Machine states, delivery and OEE KPIs, schedule health and the live schedule with a “now” line.

Adaptive scheduler Gantt scheduling board — **Scheduling board**Gantt by machine with shift calendar, maintenance, progress and late-order markers.

Adaptive scheduler optimization and approval screen — **Schedule optimization**Impact analysis and the current versus recommended schedule, ready for approval.

Adaptive scheduler what-if simulation — **What-if simulation**Preset and custom scenarios run against the live plan without changing it.

Adaptive scheduler live factory view — **Live factory**Machine and job state as the shop floor changes.

Adaptive scheduler schedule history — **Schedule history**Every schedule version with the decision behind it.

Expected outcomes

What implementing it should improve

These are the results a plant should target and measure. The scheduler reports each as a KPI, comparing the current schedule with the recommended one. We agree the baseline and targets with you in a pilot; we do not quote figures from the simulator.

More orders delivered on time

Fewer late orders and lower total tardiness, tracked as on-time delivery.

Faster response to disruption

From a machine breakdown or urgent order to a reviewed new schedule in minutes instead of a manual re-plan.

Better machine utilization

Less idle and setup time, and work moved to available machines while others are down.

Shorter overall completion time

A shorter makespan for the same set of orders.

Less planner effort

Planners review a recommendation and approve it instead of rebuilding the plan by hand.

Decisions you can explain

Each change has an impact analysis, a comparison and an audit record behind it.

With ProMonitor

Built to connect to real machine data

Today the scheduler runs on a factory simulator. Real-time data enters through a single data-provider interface, so ProMonitor’s machine states, production counts and health signals can replace the simulator without changing the scheduling engine, the approval workflow or the screens.

Status Proof of concept. The ProMonitor integration is planned; ERP and MES integration are not part of the current scope.

See ProMonitor

Technology

Python · FastAPI
Google OR-Tools CP-SAT
PostgreSQL · Redis
WebSocket live updates
Angular

Want to see it on your orders and machines?

Talk to our team about a pilot: your machines, your routings, your KPIs.

Discuss a Pilot Explore ProMonitor