Infor Makes ERP Architecture the Starting Point for Metal Fabrication AI
Infor is putting the underlying ERP architecture at the centre of its approach to artificial intelligence in metal fabrication, arguing that manufacturers need connected data and well-defined processes before they can gain meaningful value from AI agents.
In a four-part series aimed at CFOs, CIOs, COOs and CEOs in the metal fabrication industry, Infor links its CloudSuite Industrial Enterprise platform with Infor Velocity Suite and its Industry AI Agents.
The message running through the series is consistent: connect the data, understand the processes, and then introduce automation and AI agents.
Rather than treating AI as a separate technology layer, Infor positions the ERP platform as the foundation on which intelligent automation should be built.
Why ERP Architecture Matters Before AI
Manufacturers often operate complex technology environments that have developed over many years.
ERP systems may sit alongside manufacturing execution systems, product lifecycle management platforms, engineering applications, quoting tools and supplier systems. Over time, these applications can become connected through extensive integrations and custom development.
Infor identifies fragmented systems and accumulated customisation as two challenges that can make cloud modernisation and AI adoption more difficult.
Simply moving an existing environment to the cloud without addressing its underlying architecture may preserve many of the same complexities and technical debt.
Infor CloudSuite Industrial Enterprise is designed around requirements specific to industrial manufacturing. These include dimensional inventory, project costing, material pricing, configure-price-quote processes and connections between ERP, PLM and manufacturing execution systems.
By bringing more industry-specific functionality into the ERP environment, Infor aims to reduce the amount of customisation and integration required around the core platform.
That architecture also has direct implications for AI.
An AI agent working with finance, production or supply chain processes needs reliable information about orders, inventory, projects, engineering changes, suppliers and costs. If that information is spread across disconnected systems, the agent may have an incomplete picture of the process it is expected to support.
For Infor, connected and trusted operational data is therefore a prerequisite for applying AI effectively.
Process Mining Comes Before Automation
Infor’s approach also places process analysis ahead of AI deployment.
Infor Velocity Suite uses process mining to examine how business processes actually operate. The platform follows a cycle of discover, build, measure and expand, helping organisations identify bottlenecks and inefficient workflows before deciding where automation should be introduced.
This is particularly relevant when businesses are considering AI agents that can take action rather than simply provide information.
In finance, Infor highlights potential applications across areas such as project accounting, accounts payable, general ledger and accounts receivable.
Within manufacturing and operations, agents can support processes involving production orders, inventory, quality, purchasing and supplier exceptions.
Infor’s Agentic Orchestrator is designed to coordinate agents across these different processes, while the company says automated actions remain traceable to support governance and oversight.
The sequence is significant.
Instead of selecting an AI agent first and then searching for a suitable use case, Infor’s model starts with the business process. Organisations can identify where a workflow is underperforming and then determine whether automation or an AI agent can improve the outcome.
Why Metal Fabrication Is a Useful AI Test Case
Metal fabrication provides a clear example of why industry-specific ERP data matters when introducing AI.
According to Infor, raw materials can account for around 60% of total costs in the sector. This makes accurate material pricing, costing and inventory information particularly important.
Infor’s material-pricing capabilities are designed to track base and actual prices within purchasing and sales documents and apply surcharges when input costs change.
The platform also supports standard and actual costing across materials, labour and overhead, while project costing can provide visibility into expected costs and cost-to-complete before a project is finished.
These capabilities provide the type of structured information that AI agents can use to support operational and financial decisions.
The operational side of metal fabrication presents similar challenges.
Dimensional inventory may need to account for attributes such as length, width and coil diameter. Bills of materials and routing structures can also need to reflect yield, scrap and other manufacturing variables.
Infor’s approach connects ERP with manufacturing execution and engineering systems so production teams can work with current information about jobs, materials, products and quality.
For AI agents, this level of context can be critical.
A production agent needs reliable scheduling and job information. An inventory agent needs accurate visibility of material availability. A quality agent needs access to production records and non-conformance information.
The more accurately the ERP represents these industry-specific processes, the more relevant the information available to automated systems can become.
Technical Debt Becomes an AI Issue
Infor’s argument points to a wider change in how manufacturers should think about ERP technical debt.
Historically, fragmented integrations and customisations have largely been viewed as maintenance, upgrade and cost-management problems.
The emergence of agentic AI adds another dimension.
If an organisation wants AI agents to operate across production, finance, inventory or purchasing, those agents need consistent access to the underlying processes and data. Legacy integrations and disconnected systems can therefore become obstacles to automation at scale.
This means ERP modernisation decisions may increasingly need to consider not just today’s operational requirements but also the architecture required for future AI adoption.
What This Means for ERP Teams
Technical debt can now limit AI ambitions. Custom integrations and fragmented systems can make it harder for AI agents to access the data and processes they need. Addressing architecture may therefore become an important part of an AI strategy.
Process mapping should come before agent deployment. Understanding how a workflow operates today provides a stronger basis for deciding where automation belongs. It also gives technology and business teams a shared framework for measuring results.
Industry-specific ERP capabilities matter more as AI becomes autonomous. When software begins making recommendations or taking actions based on operational data, the accuracy with which an ERP represents industry processes becomes increasingly important.
Data quality remains the foundation. AI agents can only be as effective as the information available to them. Manufacturers should therefore assess data consistency, integration and governance alongside the capabilities of the AI technology itself.
The Bigger Picture
Infor’s approach reflects a broader shift in the enterprise AI conversation.
The focus is moving from simply adding AI features to ERP platforms towards creating the architecture that allows intelligent systems to operate safely and effectively.
For metal fabricators, that means AI adoption may begin well before the first agent is deployed.
It starts with understanding how the business operates, connecting the relevant data and reducing unnecessary complexity in the ERP environment.
Once that foundation is in place, AI agents can be introduced where they have the clearest opportunity to improve productivity, decision-making and operational performance.