Production scheduling software with AI: what it does for a smaller plant
Production scheduling software decides which job runs on which machine, in what order and at what time, without booking any machine or crew for more hours than it has. The AI part learns real setup and run times from your own history and builds a new plan when a machine breaks down or a rush order lands. It can also explain in plain words why an order moved. This guide is for the owner or plant manager of a smaller plant in Ontario or Quebec deciding what to buy and where to start.
What production scheduling software does
A production schedule gives every operation on every work order a machine, a start time and a finish time. The textbook version is the job shop problem. Google’s OR-Tools documentation describes it as a problem “in which multiple jobs are processed on several machines,” where each job’s tasks “must be performed in a given order” and “A machine can only work on one task at a time.” Tools such as MRPeasy show the result as a Gantt chart, one bar per job along a timeline for each machine, which a planner can adjust.
Finite capacity
Finite capacity means the schedule never books a machine for more time than it has. In Microsoft’s Dynamics 365 job scheduling, “the capacity that is scheduled can’t be larger than the capacity that’s available for the resource,” so jobs queue behind each other. Material works the same way. With finite materials, scheduling makes sure “that the required materials are available when the operation starts,” and without it, Microsoft says, “the system assumes that all items are available when they’re required.” A schedule that ignores either limit shows start times the plant cannot meet.
Sequencing and changeovers
A changeover is the time it takes to switch a machine from one job to the next: new fixtures, new tools, a new program, a first-part check. It can depend on the order of the jobs, such as moving between two materials or from a light colour to a dark one. Siemens says its Opcenter Advanced Scheduling can “Apply sequence dependent changeover times based on an operation’s attributes.” Grouping similar jobs cuts changeover hours, and the software has to weigh that against the due date of each job in the group.
Due dates and scheduling direction
Dynamics 365 schedules in two directions. Forward scheduling starts production “as early as possible,” and backward scheduling “counts backward to the latest possible date that the production can be started without missing its target deadline.” Backward scheduling holds work until it is needed, which keeps less work-in-process on the floor. Forward scheduling shows the earliest date the plant can finish.
What AI adds to production scheduling
AI improves the times that go into the schedule and how fast a new plan comes back.
Real run times learned from history
ERPs usually schedule from the setup and run standards in the routing, the list of operations for each part. Dynamics 365, for example, adjusts them by one efficiency percentage per resource: “Scheduling time = Time × 100 ÷ Efficiency percentage.” A model trained on your completed jobs can estimate times per part, machine and material, and flag standards that no longer match what the floor does. MachineMetrics, a machine monitoring vendor, says its schedule tool estimates each job’s completion from “real performance data from completed parts” and “Accounts for machine-specific variations,” and tells users to update the ERP “if actual times differ significantly.”
Re-planning after a breakdown or a rush order
When a spindle fails or a customer calls with a rush order, the plan for the rest of the week changes. MachineMetrics says its Production Schedule Intelligence “Continuously assesses and adjusts production schedules based on real-time operational data, addressing machine downtimes and delays dynamically.” Kinaxis says its scheduling can “re-optimize as conditions change.” A good re-plan shows the planner which orders moved and by how much, and the planner approves it before operators see a new sequence. When predictive maintenance flags wear ahead of a failure, the repair can go into the schedule as planned downtime.
Explaining the plan
A planner or a salesperson often needs to know why an order is late. A language model can read the schedule, the order and the constraint that moved it, and answer in plain words, for example that the order waits for a fixture in use on another job until Thursday. The answer should cite the record behind it, such as the work order and the machine calendar, so the planner can check it. AI agents for business covers what these assistants do well and where they still fail.
Start with the machine everything waits for
Tell Derik which systems your business runs on and which job slows everyone down. He will tell you what it takes to fix it.
Start a conversationProduction scheduling software vendors, and what each says it does
Scheduling comes built into some ERPs, as advanced planning and scheduling (APS) software that connects to an ERP, or as a data layer that feeds actual times into an existing schedule. The descriptions below are the vendors’ own, from their pages checked on September 28, 2026.
| Vendor and product | What the vendor says it does | Where it sits |
|---|---|---|
| Microsoft Dynamics 365 job scheduling | Schedules jobs with finite capacity and finite materials, forward or backward from a date | Inside the ERP |
| Epicor Kinetic APS | “Multiple constraint scheduling,” “visual drag-and-drop scheduling” and “real-time capable-to-promise functionality” | Module of Epicor’s ERP |
| MRPeasy | Cloud ERP “for small manufacturers (10-200 employees)” with an “interactive production calendar and Gantt charts” | Small-plant ERP |
| Siemens Opcenter Advanced Scheduling | Software that “creates production schedules based on availability of resources, constraints and materials required by the order” | Standalone APS |
| PlanetTogether APS | “Create, adjust, and execute production schedules,” connecting with systems such as SAP and NetSuite | Standalone APS |
| Kinaxis Production Scheduling | “Create realistic production schedules that optimize capacity and material usage to meet customer demand” | Supply chain planning platform |
| MachineMetrics Production Schedule Intelligence | Adjusts schedules from machine data and connects with “ERP, MES, CMMS, and APS solutions” | Shop floor data layer |
Kinaxis was founded in Ottawa and is headquartered there. Capable-to-promise means quoting a delivery date after checking capacity and material. An MES, or manufacturing execution system, records what happens on the floor, and a CMMS is maintenance management software.
What data scheduling needs from the ERP and the shop floor
From the ERP, the software needs:
- Work orders with part numbers, quantities, due dates and customer priority.
- Routings with the operations for each part, the work centre for each operation, and setup and run standards.
- Material: bills of materials, stock on hand and open purchase orders with promised dates.
- Calendars: shifts, holidays and planned maintenance for each work centre.
From the shop floor, it needs the actual start and finish of each operation, good and scrap quantities, downtime with a reason, and which machine and operator ran the job. It also needs the facts that live in people’s heads: which fixtures a job uses, which operators are certified on which machine, and which job-to-job switches take longest. A schedule also stops when material is late, and AI for inventory management and AI in procurement cover that side.
AI for ERP covers reading the ERP you already run. If yours is an older system installed on your own server with no API, legacy ERP automation explains how an agent can read its screens. Paper travellers and job tickets can be read into the record with text extraction from images.
How a smaller plant starts
- Schedule the bottleneck first. Pick the work centre that everything waits for, and sequence its jobs before anything else.
- Measure its real times. Pull a year of completed operations for that work centre and compare actual setup and run times with the routing standards.
- Try what your ERP already has. Your ERP may already include finite-capacity scheduling, as Dynamics 365 and Epicor Kinetic do. Test it before you buy a separate APS.
- Run the new plan beside the old one. For a few weeks, compare on-time delivery and changeover hours against the whiteboard or spreadsheet. AI evals explains how to test on past weeks.
- Keep the planner approving. Every re-plan goes to a named planner before it reaches operators. Human in the loop covers how to set that up.
AI strategy covers picking the first project, AI implementation covers the move to daily use, and cost of AI breaks down the spending. Finished goods leave through the warehouse and the shipping dock, which warehouse AI and AI for logistics cover. Plants that build for construction, agriculture or mining customers can read those sector guides, and the rest are in the guides.
Ask whether the schedule runs on your routing standards or on times learned from your own completed jobs, and how a planner sees what changed after each re-plan.
How ThriveAI helps
ThriveAI is an AI engineering company in Ottawa that builds private AI systems for manufacturers and distributors in Ontario and Quebec, on their own data. It brings what your ERP, email, drawings and spreadsheets hold into one database that belongs to your company. Your team asks it questions in plain language, and every answer shows where it came from. It also drafts routine work for someone on your team to approve.
ThriveAI’s systems read your ERP as it is, including older versions installed on your own server, and every connection only reads data. The platform is designed to keep each client’s data on its own server in Canada. You choose a model on that server or a hosted model under a written zero data retention agreement, and a hosted model may process requests outside Canada. Derik Lawlis, the founder, leads every project and stays close to the build. Stage one is a working tool for the one job that costs you the most, at a fixed price, in weeks. About ThriveAI covers the company.