Short answer
The best way to do production scheduling with AI in a job shop is to have AI draft the weekly schedule from your open jobs, due dates, routings, work-cell capacity, labor, and material status, then let a planner review and approve it. The AI applies rules your shop defines and flags conflicts and jobs that will ship late. When the floor reports progress, the schedule updates. Lattice Partners is an AI implementation partner based in Los Angeles and New York that helps companies build an AI strategy, trains their teams on AI tools, and builds custom agents and software, including scheduling tools for fabricators.
- Many job shops still build the weekly schedule in Excel, and most of the time goes into chasing people for inputs.
- You can buy an ERP scheduling module, buy dedicated APS (advanced planning and scheduling) software, or build a tool you own on top of your ERP.
Why does job shop scheduling take so long today?
Job shop scheduling takes hours each week because the inputs are scattered and change daily. A planner opens last week's spreadsheet, then calls or walks the floor to learn which jobs moved, whether material arrived, who's out, and which machine is down. The spreadsheet holds the answer only as long as nobody changes anything.
Fabricators feel this more than repetitive manufacturers, because every job has a different routing and engineering changes land mid-week. The schedule ends up living in one person's head.
How does AI production scheduling work in a job shop?
AI production scheduling works by pulling the inputs together automatically and proposing a schedule against your rules, with the problems marked so the planner can fix them before making the final call.
In a typical week, the cycle runs in five steps.
- Pull inputs. The tool reads open jobs, due dates, and routings from the ERP (enterprise resource planning) system, labor availability from timekeeping, and material status from purchasing and receiving.
- Propose a schedule. The tool sequences operations through each work cell. ERPs already support forward scheduling (start date in, finish date out) and backward scheduling (due date in, start date out), as Oracle's NetSuite documentation describes. The AI layer adds your judgment rules on top.
- Flag problems. Overloaded work cells, jobs that can't make their due date, operations waiting on material, and crews scheduled past their hours all get called out.
- Planner reviews. The planner moves jobs or overrides priorities, then approves. The tool records the reason for each change so the rules can improve.
- Update from the floor. As operations complete or slip, percent complete and delivery forecasts recalculate.
| Input | Where it comes from | What the AI does with it |
|---|---|---|
| Open jobs and due dates | ERP or job tracking spreadsheet | Ranks jobs by due date and priority rules |
| Routings and estimated hours | ERP routings or the original quote | Breaks each job into operations per work cell |
| Machine and work-cell capacity | Shop calendar, maintenance log | Loads each cell and flags overloads |
| Labor availability | Timekeeping, PTO calendar, crew skills list | Matches operations to qualified people on shift |
| Material status | Purchase orders, receiving records | Holds operations until material is on hand, flags late POs |
| Shop floor status | Tablets, scan points, supervisor updates | Updates percent complete and reforecasts ship dates |
| Shop rules | Written with the planner | Applies setup grouping, rush handling, customer priority |
What rules should the shop define?
Start with the scheduling rules a good planner already follows in their head. Writing those down is often the most useful part of the project, since the AI can't propose a schedule the planner would sign off on without them. Common ones:
- Which customers or job types jump the queue
- Grouping jobs by material or setup to cut changeovers
- Which welders are certified for which processes
- Maximum overtime per person per week
- Buffer time before ship dates for paint, galvanizing, or inspection
- Which work cells can substitute for each other
Some rules won't survive being written down. The planner handles those in review, and the tool logs each override.
"The first part of a scheduling project is writing down the rules that live in your planner's head. Once those are written down, the AI can do the math."
Martin Anquetil, Co-Founder and Managing Partner, Lattice Partners
What are the options for production scheduling software for fabricators?
Fabricators have three main options for production scheduling software: an ERP scheduling module, dedicated APS software, or a custom tool built on top of the ERP they already run. Which one fits depends mostly on how clean your ERP data is and how unusual your rules are.
| Option | How it's priced | Fits best when | Watch out for |
|---|---|---|---|
| ERP scheduling module | Often an add-on to a yearly ERP subscription | Your ERP data is clean and your process matches the module | Rigid for custom work; another recurring license |
| Dedicated APS software | Usually a separate subscription | You run many machines and need finite-capacity math | Integration work; your team learns another system |
| Custom tool on your ERP | One-time build you own | Your rules are specific and you want a single view | Needs a builder who knows shop operations |
On ERP pricing, NetSuite is a useful example. Its own site says customers "subscribe to NetSuite for an annual license fee" made up of "core platform, optional modules and the number of users." Other ERPs used by fabricators, such as Epicor, Acumatica, and JobBOSS, have their own packaging, so check what scheduling features your current license includes before buying anything new.
Quoting uses much of the same data. See the best way to do quoting with AI for a fab shop.
How does scheduling connect to job tracking and a big board?
The same data drives scheduling and job tracking, so the two connect naturally. Once jobs, operations, and floor status sit in one place, it's much easier to build a shop-wide "big board" that follows every job from material receipt to shipment and shows which work cell it's in and how many hours it has used against budget.
Plenty of shops start with the big board, since it ends the "where's my job?" calls before anyone touches how the schedule gets built. Scheduling gets added later on the same data. Actual hours from timesheets feed back into both, which also sharpens future quotes.
For the drawing side, see using AI to read drawings and estimate fabrication jobs.
How Lattice Partners does this
Lattice Partners builds custom scheduling and job tracking tools for metal fabricators and job shops. The tool connects to your ERP, timekeeping, and purchasing records and drafts the schedule against rules we write down with your planner. The floor and the office both work from the same view of job status.
We start by sitting with the planner to learn how the schedule gets made today and which rules matter. The tool gets tailored to your work cells and crews. We aim to ship a first version within a few weeks, then improve it from real weeks of use over 2 to 3 iterations. It's a one-time build. You own the code, with no per-seat or per-module license from us. Cost depends on scope.
Frequently asked questions
Can AI fully automate production scheduling in a job shop?
AI can draft the full schedule, but a planner should approve it. Rush orders and surprises on the floor need human judgment. Where AI saves time is the hours a planner spends gathering inputs and catching conflicts.
Do we need an ERP to use AI production scheduling?
AI production scheduling works best with an ERP, but it doesn't require one. A tool can read jobs and routings from well-kept spreadsheets, though inconsistent spreadsheets will show up as a worse schedule.
What's the difference between AI scheduling and APS software?
APS software focuses on finite-capacity math across machines and orders. AI scheduling adds reading messy inputs, applying written shop rules, explaining conflicts in plain language, and learning from planner overrides. Some shops use both.
How does AI capacity planning handle labor?
AI capacity planning for manufacturing handles labor by matching each operation to people who are qualified and on shift. It reads PTO calendars and skills lists, then flags operations that have no available person. Overtime limits come from the rules the shop defines.
Related reading
About Lattice Partners
Lattice Partners is an AI implementation partner based in Los Angeles and New York. We help companies build an AI strategy, train their employees on AI tools like ChatGPT and Claude, and build custom AI agents and software. Clients own the code we build. Lattice Partners was founded by Colin McDonnell, Martin Anquetil, and Christian Yeghnazar. Contact us to talk about your use case.