Artificial intelligence can streamline. Improve efficiency. Free up your calendar by letting AI take care of the tedious, time-consuming reporting. From automating routine tasks to supporting faster decisions, the opportunities are significant.

But there is a problem with the way many businesses approach AI: they focus on what the technology can do before considering whether their business is ready to use it.

Better automation starts with better processes

Imagine introducing an AI agent to automate part of your order management, purchasing or production planning. The potential benefits are clear: less manual work, faster processes and fewer repetitive tasks.

But what happens if the underlying data is incomplete? What if different departments use different definitions? What if the business rules are unclear or the process itself is inconsistent?

The AI may execute tasks faster, but that does not necessarily mean it will produce better outcomes. Automation amplifies the processes it relies on. When those processes are well defined, automation can create significant value. When they’re not, AI can amplify existing problems rather than solve them.

Before asking what AI can automate, ask whether the process is ready to be automated.

Start with your ERP foundation

Your ERP system plays a central role in how your business operates. It connects information and processes across areas such as purchasing, inventory, production, sales and finance. In many ways, ERP is the brain of the company.

But the value of an ERP system is not determined by the amount of data it contains. It also depends on the quality, consistency and structure of that data, and on how well the system reflects the way the business actually works.

There are three areas worth examining before scaling AI initiatives.

1. Data quality and definitions

AI depends on information it can interpret and use reliably. If data is incomplete, outdated or inconsistent, the results can be unreliable too. Businesses need clear definitions, appropriate data ownership and processes that maintain data quality over time. This is not simply a technical exercise. Data quality depends on the people and processes responsible for creating, maintaining and using information throughout the organisation.

2. Core processes and business rules

Not every business process needs to be identical. But businesses need to understand which processes are fundamental, which rules govern them and where flexibility is appropriate. A strong ERP foundation provides structure around these core processes.

This is particularly important in manufacturing, where production, materials, traceability and quality requirements are closely connected. An ERP system designed for the realities of the industry can provide more of the relevant business logic from the outset. That gives businesses a clearer starting point for automation.

3. Governance and control

As AI becomes more involved in business operations, organisations need to understand how it accesses information, how automated actions are governed and where human oversight is required. Requirements vary by business and industry, but the principle is the same: automation should not come at the expense of visibility and control.

Businesses need appropriate safeguards, clear responsibilities and an understanding of how their data moves between systems. AI readiness is therefore about more than adopting new technology. It is also about establishing the conditions for using that technology responsibly.

From a solid foundation to smarter automation

Getting the foundations right does not mean postponing AI until every process is perfect. It means starting with the right opportunities and building from there.

A practical approach could look like this:

Step 1: Identify a real business problem.
Start with a process that consumes time, creates friction or requires unnecessary manual work. Avoid introducing AI simply because the technology is available.

Step 2: Understand the process.
Map the relevant steps, business rules, data sources and exceptions. Identify where the process works well and where problems occur.

Step 3: Check the data.
Determine whether the information required for the use case is accessible, sufficiently reliable and consistently defined.

Step 4: Establish the right controls.
Decide what the AI solution should be allowed to do, how its output will be checked and when a person needs to intervene.

Step 5: Test, learn and expand.
Start with a defined use case, evaluate the results and improve the solution before extending it to other processes.

This approach helps businesses connect AI initiatives to measurable operational value rather than treating AI as a separate technology experiment.

Open systems create opportunities. Good architecture makes them manageable

AI solutions do not operate in isolation. They need access to relevant information and, in some cases, the ability to interact with business systems. Technologies such as APIs and MCP can help make these interactions possible. They create opportunities to connect ERP data with AI tools, agents and other applications.

But connectivity alone is not the goal. As the number of integrations and automated workflows grows, businesses also need to manage how information moves between systems, monitor integrations and maintain appropriate control.

An open architecture can create flexibility, but that flexibility needs to be supported by a structure that makes the overall environment manageable. The objective is not simply to connect more systems. It is to create a connected environment in which data can be used effectively and responsibly.

AI readiness is an operational question

It is tempting to view AI readiness as a question of technology: Do we have the right tools? Do we have access to the latest models? Can we build an agent?

Those questions matter. But they are only part of the picture.

A more useful starting point is to ask:

    • Do we trust the data our business relies on?
    • Are our core processes clearly defined?
    • Do we understand the rules that govern our operations?
    • Can our systems exchange information in a structured and manageable way?
    • Do we have appropriate oversight of automated decisions and actions?

The answers will help determine where AI can create value today and what needs to improve before it can be scaled.

Build the foundation. Then unlock the possibilities.

AI will create new opportunities for manufacturers and other complex businesses. It can help reduce manual work, improve access to information and support new ways of managing operations. But the strongest results are unlikely to come from AI alone.

They will come from combining the right technology with reliable data, sound processes and a clear understanding of how the business works.

AI won't fix your ERP foundation. But a strong ERP foundation can help you get far more value from AI.

The question is not simply how quickly you can introduce AI. It is whether your business is ready to turn its potential into lasting operational value.