ERP AI Readiness: Preparing Business Central for AI
Imagine hiring an incredibly capable new employee who can work around the clock, analyze information in seconds, and take action without being reminded.
Now imagine that employee is an AI agent. That's when ERP AI Readiness gets a lot less theoretical.
Now let’s say we give this AI agent access to an ERP full of duplicate customer records, inconsistent processes, mystery permissions, and workflows held together by “Ask Steve. He can explain.”
That’s where ERP AI Readiness becomes important.
In the first article in this series, I explained what agentic AI means for ERP. In the second, we looked at how agentic ERP workflows could change the way everyday work gets done. (I’ve included links to both at the end if you’d like to catch up.)
Now comes the less glamorous (but arguably more important) part: making sure the environment those agents will work in is ready for them.
If you’re already using Microsoft Dynamics 365 Business Central, you may be further along than you think. You don’t need perfect data, flawless processes, or a six-month “AI transformation initiative” before you can start preparing.
In this final article of the series, I’ll explain how to evaluate your data, workflows, security, permissions, and governance, and what Business Central users can start doing today to build an ERP foundation that’s ready for what comes next.
What does ERP AI readiness look like?
Your ERP doesn't have to be perfect before you start preparing for agentic AI. But an AI agent shouldn't have to sort through conflicting data and inconsistent processes to figure out how your business is supposed to work.
I’d start an ERP AI Readiness assessment with three questions:
Is the data reliable?
Are the workflows reasonably consistent?
Are there clear rules about what an AI agent can and can't do?
Then look for the places where employees regularly work around the ERP rather than through it. A spreadsheet that fills a legitimate gap is one thing. Twelve spreadsheets called "FINAL_inventory_v7" may be trying to tell you something.
And if the explanation for a critical process ends with “Steve can explain,” that’s worth documenting too. Steve deserves a vacation.
The goal isn't to make your ERP flawless before AI enters the picture. It's to understand where the foundation is solid, where it needs attention, and where giving an AI agent more autonomy could magnify a problem instead of solving one.
What data problems prevent AI agents from working effectively?
AI agents depend on the data they can access, so inaccurate, incomplete, duplicated, or inconsistent information can quickly undermine the results they produce.
Think about the first week: On Monday, the ERP says Customer A gets a 10% discount.
On Tuesday, another record says it's 15%. By Wednesday, the AI agent has identified three versions of the same customer and Steve is being summoned to explain which one is right.
AI isn't the problem.
That's one reason ERP data governance matters so much when you're preparing for agentic AI.
Deloitte found that nearly 70% of manufacturers surveyed identified data-related challenges—including quality, contextualization, and validation—as their biggest obstacle to implementing AI.
Before asking AI agents to do more, look for duplicate records, inconsistent naming conventions, missing fields, outdated information, and important data that still lives outside the ERP. Your data doesn't need to win any awards for cleanliness.
But it does need to be reliable enough for an AI agent to know which version of the truth to trust.
Clean, structured, reliable data doesn't just improve reporting today. It's one of the foundations of an AI-ready ERP.
How should ERP permissions change when AI agents can take action?
When AI agents can take action inside business systems, permissions need to reflect exactly what those agents are responsible for doing. And nothing more.
You probably wouldn't give your AI agent unrestricted access to every financial record, customer account, approval, and system on its first day simply because of its capabilites. AI agents need the same boundaries as human employees.
Start with the principle of least privilege: give an agent access only to the data and actions necessary for its role. Clearly define which decisions it can make, which actions require human approval, and where it should stop and hand something over to a person.
Microsoft's guidance for Copilot Studio similarly emphasizes governance around authentication, actions, connectors, data policies, and auditing. As agent capabilities expand, those controls become increasingly important.
This is also a good time to review existing user roles and permissions. If Steve still has administrator access because somebody checked a box during an implementation eight years ago, ERP security best practices probably deserve some attention before you give an AI agent access to the system.
Good governance doesn't make an AI agent less useful. It gives the agent a safe lane in which to be useful.
Do I need to redesign my business processes before adopting agentic AI?
No. You don't need to redesign every business process before adopting agentic AI. You do need to understand the important ones well enough to explain how they're supposed to work.
That's a meaningful distinction.
Start with the Business Central workflows where an AI agent could eventually play a role: approvals, purchasing, order processing, inventory management, reporting, and exception handling.
Microsoft Power Automate provides tools for mapping and automating business processes, but before you automate anything, it's worth asking whether the process itself is consistent.
If one manager approves a purchase at $5,000, another at $10,000, and Steve says, “It depends,” even the best AI agent is going to have trouble figuring out the rules.
Document the core steps. Clarify who owns the process. Identify unnecessary handoffs and the exceptions that genuinely require human judgment. Business Central optimization can start with improvements like these rather than a sweeping redesign.
The objective isn't process perfection. It's making sure your AI agent can follow the rules without needing Steve on speed dial.
What should Business Central users do today to prepare for agentic AI?
Start small. ERP AI Readiness doesn't require a massive transformation project. For Business Central users, it can begin with five practical steps:
Review data quality. Identify duplicates, missing information, and inconsistent master data.
Standardize core workflows. Focus first on processes where exceptions and workarounds are common.
Review governance and security. Make sure roles, permissions, and ownership still reflect how your business operates today.
Identify manual bottlenecks. Look for repetitive tasks, unnecessary handoffs, and places where “Steve can explain” has become part of the process.
Build an AI readiness roadmap. Prioritize the improvements that will make your ERP more useful today while preparing it for greater AI capabilities tomorrow.
You don't have to get everything AI-ready at once. In fact, many of these steps are simply good Business Central optimization. The difference is that you're now looking at them with AI agents in mind.
ERP AI Readiness isn't about predicting exactly what AI agents will be able to do next year. It's about building an ERP foundation you can trust as those capabilities evolve.
Clean up the data. Clarify the processes. Tighten permissions. Document the things only Steve knows. The goal isn't to replace Steve with AI.
It's to stop making Steve the company's unofficial operating manual.
If you'd like to catch up on this series, start with Agentic AI in ERP: What Business Leaders should Know, then explore How Agentic ERP Workflows Keep Work Moving.
And if you're wondering where your Business Central environment stands today, reach out. We can help you figure out what makes sense to tackle first.
About Matt Keyes
Matt Keyes is a visionary leader, founder, and CTO of Key Partner Solutions. With over two decades of experience in Microsoft Dynamics, he is passionate about driving digital transformation for businesses through innovative technology solutions.
His deep technical expertise, combined with a strategic approach to solving business challenges, makes him a sought-after thought leader in the industry.
Today, Matt is focused on empowering companies to unlock new levels of growth and efficiency through cutting-edge software development and consulting.
Connect with Matt on LinkedIn.
