Business Central + fabrication AI
AI for Dynamics 365 Business Central in Metal Fabrication
Short answer
AI for Business Central in metal fabrication reads items, variants, production orders, routings, subcontracting and lot records so estimators, schedulers and quality staff can ask about job cost, heat-lot traceability and outside-processing status in plain English. A private model grounds answers in your data and shows the query, while quoting and write-backs stay as drafts for approval.
- ERP
- Dynamics 365 Business Central
- Industries
- Metal Fabrication, Job Shops, Sheet Metal and Welding
- Written for
- Operations Manager
Fabricators and job shops running Business Central usually live with a particular tension: the ERP models production orders, routings and BOMs well, but a shop's real world is sheet and plate in odd sizes, remnants, cut lists, welding and finishing operations, outside processing at galvanizers or powder coaters, and customers who want material certifications by heat number.
Much of this is handled through conventions. Material grades and thicknesses become items or variants, heat numbers and mill certificate references become lot numbers, outside processes use work centers flagged as subcontracting and the Subcontracting Worksheet, and nesting happens in a separate CAM tool whose output arrives as a spreadsheet or a custom import. The conventions work, but they live in people's heads.
That is where grounded AI is useful: reading the real posted data to answer how last quarter's jobs compared with their estimates, where a heat lot went, which jobs are waiting on a vendor, and what remnant stock could cover a new request. It is not a nesting engine and it does not weld anything; it shortens the time between a question and an evidence-backed answer.
This page describes what that looks like on Business Central, which objects and integration points matter, how to protect customer drawings and quoted pricing, and when a simpler tool or a specialist add-on is the better choice.
What usually gets in the way
The problems we hear most from operations manager teams running Dynamics 365 Business Central.
Estimates and actuals are never compared
Quoted hours and material sit on the sales quote or project; actual hours accrue in capacity ledger entries and material in consumption. Comparing by part family or operation rarely happens, so the same underestimates repeat.
Remnants and offcuts disappear
Leftover sheet and bar often exist only as a bin location or a lot with an odd quantity. Estimators order new stock because finding a suitable remnant is not practical.
Heat-lot certs must follow the part
Customers in structural, pressure and aerospace-adjacent work ask for mill certificates by heat number. Linking a finished weldment back to every heat consumed is manual unless tracking was done consistently.
Outside processing is a status black box
Parts sent for plating, galvanizing or heat treat sit on subcontract purchase orders. Schedulers chase vendors by phone because the system shows a due date but no real status.
Price volatility in steel and aluminum
Surcharges and material price swings between quote and purchase affect margin. Spotting which open quotes are exposed means recalculating cost against current vendor prices.
Where AI earns its place in Dynamics 365 Business Central
Each use case names the ERP objects it reads or writes, so your ERP team can judge the integration effort before anyone commits budget.
Estimate versus actual job analysis
Estimators ask how a part family performed against its quoted hours and material. The assistant compares routing time, capacity ledger actuals and consumption by job, and highlights operations that consistently overrun.
Touches: Production Order, Routing Line, Capacity Ledger Entry, Item Ledger Entry, Sales Quote Line
Outcome: Evidence for adjusting standard times and quote factors; estimators still decide the changes.
Heat-lot traceability by shipment
For a customer's shipment, the agent lists the heat lots consumed in each part, with certificate document references, ready for the quality lead to assemble the package.
Touches: Item Ledger Entry, Lot No. Information, Item Tracking Line, Sales Shipment Line, Document Attachment
Outcome: Cuts the time to compile certification packages and exposes parts consumed without a lot recorded.
Remnant and offcut matching
Given a new job, the assistant searches remnant stock by material, thickness, and minimum dimensions stored in variants or custom fields, and suggests candidates before purchasing new sheet.
Touches: Item, Item Variant, Bin Content, Lot No. Information, Production Order Component
Outcome: Less purchased material for small jobs where suitable offcuts already exist.
Outside processing tracker
The agent lists subcontract purchase orders by vendor, expected return date and the production orders they gate, and drafts follow-up emails to vendors for the buyer to send.
Touches: Purchase Line, Work Center (subcontracting), Subcontracting Worksheet, Prod. Order Routing Line, Vendor
Outcome: Replaces phone-tag scanning with a short list of at-risk jobs.
Quote exposure to material price changes
The assistant recomputes material cost on open quotes against latest vendor prices and flags those whose margin now falls below a threshold.
Touches: Sales Header, Sales Line, Purchase Price, Item Vendor, Production BOM Line
Outcome: Earlier repricing conversations on exposed quotes.
Scheduling bottleneck summary
Daily, the agent summarizes load on laser, brake, weld and paint work centers against capacity and lists jobs likely to miss ship dates.
Touches: Work Center, Work Center Calendar Entry, Prod. Order Routing Line, Sales Line
Outcome: A clearer morning view for the production meeting; the schedule is still set by people.
Nonconformance and rework lookup
Quality staff ask for rework history by customer, part and weld cell; the assistant reads scrap and rework postings and linked comments.
Touches: Output Journal (scrap), Prod. Order Line, Item Ledger Entry, Capacity Ledger Entry
Outcome: Faster root-cause conversations backed by the posted history.
Reference architecture
The AI layer reads Business Central production, inventory and purchasing data through scoped API pages and indexes drawings, certificates and WPS documents on your infrastructure. It does not alter nesting or CAM output; it connects what posted in the ERP to what people ask about it.
- 1
ERP data access
Read-only endpoints for production orders, routing and capacity ledger entries, item ledger, lot records and purchase lines.
- 2
Document index
Mill certs, drawings, weld procedure specs and inspection reports indexed locally with item, job and lot metadata.
- 3
Shop conventions map
How your shop encodes thickness, grade, remnant dimensions and heat numbers, written down so the model does not guess.
- 4
Model serving
Open-weight models running on your GPUs or private cloud, keeping customer drawings and prices inside your boundary.
- 5
Approval layer
Drafted emails and proposed quote changes are queued for staff; no autonomous postings.
Integration notes for your ERP team
- Standardize how heat numbers map to Lot No. and ensure Item Tracking Codes are on for structural and certified materials.
- Subcontracting uses work centers with a subcontractor and the Subcontracting Worksheet; expose the resulting purchase lines linked to the production order.
- Remnant dimensions need a home: item variants, lot attributes or custom fields. Without it the AI cannot match offcuts.
- Nesting software output arrives via spreadsheet or custom import; record the nest identifier on the order so jobs can be tied to yield.
- Index certificates and drawings from file shares or SharePoint with job, item and lot IDs in metadata.
- Cache master data like routings and work centers; keep ledger queries date-bounded.
Deployment options
Small on-prem GPU server
Shops with controlled customer drawings or ITAR-related work
A single server runs the model and index inside the shop network, reading the tenant or on-prem database via scoped access.
Private cloud tenant alongside SaaS
Growing fabricators on Business Central online
Model runs in your Azure subscription; service-to-service OAuth reads only the objects exposed.
Shared service for multiple shops
Groups with several plants
One hosting location with per-entity access controls and separate indexes per company.
Compliance and data control
How the architecture supports your obligations. Certification and accountability stay with your organisation; the design keeps the evidence straightforward.
EN 1090 and ISO 3834 where applicable
Traceability and weld documentation retrieval is faster; certification and procedures remain the shop's responsibility.
AWS D1.1 and customer weld requirements
Index procedure documents and welder records so questions retrieve them with references; qualification decisions stay with certified staff.
ITAR/EAR for controlled drawings
Run in a controlled, access-limited environment so technical data is not sent to public AI services; customer responsibility for registration and controls.
AS9100 and customer quality flow-downs
Lot traceability and rework history are retrievable with sources for audits.
Pricing confidentiality
Quotes and vendor prices stay in your infrastructure with role-based access.
Where Netray fits
How an engagement runs
Phase 1 . 1 to 2 weeks
Shop conventions review
- -Documented material, remnant and heat-lot conventions
- -Sample questions from estimators and quality
- -Access plan
Phase 2 . 2 to 3 weeks
Data and document layer
- -Read-only API pages
- -Indexed certificates and drawings
- -Test question set
Phase 3 . 3 to 4 weeks
Pilot with estimating and scheduling
- -Estimate-versus-actual assistant
- -Outside processing tracker
- -Accuracy review
Phase 4 . 2 to 4 weeks
Extend
- -Remnant matching
- -Quote exposure alerts
- -Operations runbook
Questions to ask any vendor, including us
A short list that separates real Dynamics 365 Business Central AI work from a chatbot demo.
- How do we record heat numbers and remnant dimensions so the AI can use them?
- Does the assistant touch our nesting software or only read posted ERP data?
- Where do customer drawings and pricing get processed?
- How are estimates compared with actuals across revisions of routings?
- What does it write, if anything, and who approves it?
- How do you handle controlled technical data?
Frequently asked questions
Can AI do the nesting or cut optimization?
No. Nesting is a specialized optimization problem handled by CAM software. The AI helps around it: recording yields, comparing nested versus quoted material, matching remnants and explaining overruns. If you need better nesting, evaluate nesting tools rather than a language model.
Is Business Central good enough for job shops?
Many fabricators run it successfully, particularly with a few extensions for sheet-metal conventions, but native support for remnants, nesting and finishing steps is limited. Some choose ISV add-ons. AI is not a substitute for choosing the right foundation; it adds value once the data is posted consistently.
How does heat-lot traceability work?
Business Central tracks lots on items with an Item Tracking Code. If you receive plate or bar by heat number as a lot and consume lots into production, the ledger holds the link. The assistant traces from shipment to heat lots using those entries and attached certificates, and reports gaps where lots were not recorded.
Will it improve our estimates?
It gives estimators evidence by comparing quoted and actual hours and material per part family and operation. Improvement depends on acting on that evidence by adjusting routing times and factors. Where actual time posting is sloppy, the first outcome is a data quality finding.
Can it chase vendors for us?
It can draft follow-up emails for subcontract purchase orders that are late or at risk, listing the jobs affected. A buyer reviews and sends them. We do not recommend autonomous vendor contact, because relationships and negotiated exceptions matter.
Is controlled drawing data safe?
The design keeps models and indexes on your servers or in a private environment you operate, with open-weight models and local logging, so drawings are not sent to public AI services. Your own export control program and access policies still govern who may use the system.
What size shop is this practical for?
Shops with a few thousand jobs a year and a few dozen users tend to find value, as the volume of data and recurring questions justify a layer. Very small shops may get most benefit from Power BI and good routing discipline, and we would say so after a short review.
Related guides
AI for Business Central manufacturing operations, beyond Copilot
AI for Business Central manufacturing beyond Copilot: grounded answers on production orders, routings, and capacity via APIs v2.0, on-prem capable.
Business Central + private AIAI for Dynamics 365 Business Central, on-premises or SaaS, beyond Copilot
AI for Business Central beyond Copilot: grounded question answering over BC's data via OData and APIs, for on-premises and SaaS tenants alike.
AS9100D + on-prem AIAI for AS9100 Quality Management on Your ERP
AI on top of your ERP quality module for AS9100D suppliers: NCR/CAPA drafting, FAI support, counterfeit parts screening, with a full audit trail.
ProShop + on-prem AIAI for ProShop ERP in AS9100 Machine Shops
Add AI on top of ProShop ERP to speed up NCR/CAPA drafting, FAI review, and document control, without moving AS9100 quality records off your network.
PO follow-up that does not wait on a personAI Purchase Order Automation: Supplier Follow-Up and Expedite Without a Buyer Chasing Email
AI drafts routine PO confirmation and expedite messages from open ERP order data, with a buyer approving every message and change before it goes out.
ITAR + on-prem AIITAR-Compliant AI for ERP Technical Data
How to add generative AI to your ERP without creating a deemed export under ITAR. On-prem architecture patterns an Empowered Official can sign off on.
Plan it with numbers
Scrap Rate Cost Calculator
Turn your scrap percentage into an annual dollar figure - including labor and overhead absorbed by scrapped units - and see what each point of improvement is worth.
Free ToolFirst Article Inspection (AS9102) Checklist
A 30-point checklist for complete, first-pass-acceptable AS9102 FAIs - planning, Form 1/2/3 completeness, characteristic accountability, and the delta FAI triggers teams miss.
Free ToolCycle Time & Capacity Calculator
Translate cycle time, parallel stations, shift pattern, and realistic utilization into hourly, daily, and annual production capacity.
GuideFinite Capacity Scheduling: A Practical Guide
Finite capacity scheduling explained: how it differs from infinite MRP loading, APS options for SyteLine and LN, data prerequisites, and AI-assisted scheduling.
GuideAn ERP Data Quality Framework
An ERP data quality framework for manufacturers: profiling item and BOM data, six quality dimensions, automated rules, scorecards, and remediation workflows.
GuideFrom AI Pilot to Production: The Playbook
Move an AI pilot to production: environment promotion, monitoring, rollback plans, and the ERP integration hardening manufacturers need before go-live.
Talk it through with an engineer who knows Dynamics 365 Business Central
Bring one real question your team cannot answer from the ERP today. We will map the data path, the model, and where it runs, and tell you honestly if AI is the wrong tool for it.