How to Prepare Your Business for AI-Powered ERP

Since the release of the Copilot capabilities in Microsoft Dynamics 365 Business Central, or any other AI in any other business, naturally there has been a lot of chatter about its usefulness and what it can or cannot do.

Some of the feedback is enthusiastic, some are skeptical, and some are frustrated. That’s expected, especially when new technology arrives. Software vendors bring hype and excitement, while users bring a healthy dose of doubt.

What we know for sure is that the ones who embrace change, which is new technology in most cases, will be the winners of long-term value creation by way of increasing productivity and the quality of their work. The challenge is that this is coming at us at full speed, and most of us don’t even know where to begin using it the right way in our personal lives and workplaces.

So how do we get on this beast coming at us at a hundred miles an hour, which is not looking to slow down any time soon? What if we don’t make the leap in a timely manner? How do we prepare for such a leap? When and where do we even begin utilizing AI and Copilot and all its tentacles infused in all Microsoft applications, including Business Central?

Short answer: Don’t rush but start now. Here is why and how.

Learning to use AI to our advantage is no different from learning anything else. This time around, the stakes are higher. Not only does a business’s success depend on adoption rate, but it’s also not a clear-cut A-or-B choice. So, better preparation matters more than ever at this critical junction and in the never-ending cycle of change management.

The path to success goes through educating both the leaders and the users. Building excitement boosts motivation and turns anxiety into positive energy, but it needs to be backed by intentional education, a clear process, and the awareness that we all need to row in the same direction in harmony. Otherwise, we’ll be spinning around exactly where we stand today with a lot of energy and money spent and not go anywhere.

 

How Should You Get Started With Copilot?

Stay positive, and remember, it’ll take some trial and error to get used to an AI model.

AI models appear to follow the business values and ethical standards of their creators. In my experience, which is the case with most long-term useful products and life hacks, moderation, discipline, and modesty benefit the end user in the long run.

Why do I say that? Because I’ve seen Microsoft not be the best at a lot of things over the past thirty years of my professional life, but it’s still the dominant toolbox I go to when I need to get work done.

And Copilot is no different. It is not the most intelligent or the most cavalier of the LLMs, but that’s because we don’t want it to be. We prefer security, governance, and a roadmap over rogue agents we can’t trust.

Once we are ready to welcome Copilot into our lives, take it slow. Do not turn on every Copilot feature and hope for the best. Evaluate where your organization is today, including your most valued resource: The People. Just like any other project, identify who could be the product champions; energized to learn, be okay to make mistakes (leadership support is very important here), mitigate risk, and most importantly, be willing to teach others.

Then define the desired business outcomes and have flexibility. Because with AI, one thing is certain: What is true today may look different just a short while from now.

 

How Useful Is Microsoft Copilot?

The honest answer is that it depends.

Copilot is not a magic wand and treating it like an old software switch that simply gets turned on will likely lead to disappointment. It will not automatically correct inconsistent processes, clean up poor data, resolve overly broad security access, or document procedures that may exist only in experienced employees’ minds. In many cases, AI will expose those weaknesses rather than conceal them.

That does not mean Copilot is ineffective. It means expectations need to match reality. Its usefulness depends partly on the business environment it operates in and the problem it is being asked to solve.

A well-defined process supported by reliable data and clear decision rules gives Copilot and AI agents a much better opportunity to deliver value. An inconsistent process built on tribal knowledge gives them very little to work with.

Microsoft gives Business Central administrators controls to review available Copilot and agent capabilities, deactivate individual capabilities, manage user access, and apply data governance controls. That is an important reminder that successful adoption involves more than testing features. It also requires deliberate decisions about access, control, and responsible use.

The better question, then, is not simply, “How good is Copilot?” It is, “How prepared are we to use it well?”

 

What Should You Do Before Automating a Business Process?

Start With the Process, Not the Agent

Expense management provides a useful example. It is easy to see the appeal of an agent that can read a receipt, extract the merchant, date, and amount, and create an expense entry. That eliminates tedious work and improves employee experience. But creating the expense entry is only the beginning.

The organization still must decide how to review, approve, reimburse, and record that expense. A meal may follow different rules from mileage. A project-related expense may need to be assigned to a customer or a job. A charge above a certain amount may require another level of approval. A policy violation may generate a warning, block submission, or be routed to a manager for review.

AI should not answer those questions on behalf of the business. They are business decisions that need to be made before the technology can execute them consistently.

This is why process preparation matters. If expense policies are documented, approval responsibilities are clear, and exceptions are handled consistently, an agent can help move the work through the organization more efficiently. If every manager handles expenses differently, automating receipt capture solves only the first few minutes of a much larger process.

The same principle applies to sales orders, payables, bank reconciliation, reporting, and nearly every other AI opportunity. Before asking what Copilot can automate, ask whether the organization agrees on how the work should be done.

 

How Can You Encourage AI Experimentation Without Creating Islands?

Most companies will begin their AI journey informally, and that is not necessarily a bad thing. Someone in finance may experiment with reporting. A salesperson may use Copilot to summarize customer information. An operations employee may test an agent against a repetitive task.

These early users often become product champions who help the rest of the organization understand what is practical and where limitations remain. The problem begins when experimentation never develops into coordination.

If every department creates its own approach, finance may establish one set of practices while sales, operations, and customer service establish entirely different ones. The organization can wind up with disconnected processes, inconsistent expectations, duplicated effort, and no common understanding of when human review is required.

The answer is not to suppress experimentation or build an enormous governance committee before anyone is allowed to learn. It is to place sensible guardrails around that experimentation.

Leadership should identify initial use cases, assign responsible owners, define where human review is required, document what is learned, and give successful ideas a path to broader adoption.

Individual curiosity can start the journey, but organizational alignment is what allows it to scale.

 

The Best AI Strategy Does Not Start With AI

It starts with the business and the people who make it work. Copilot, agents, and AI-powered workflows are tools. Their value depends on clear objectives, strong processes, reliable data, a strong governance model, and employees willing to learn.

That is why preparedness should not be confused with hesitation. Preparing does not mean forming a committee that discusses AI for a year while competitors learn by doing. It means creating enough structure to begin responsibly, learn quickly, and adjust intelligently.

At OTT, we help organizations examine real business processes, identify practical opportunities for improvement, and determine where Business Central, Copilot, and AI-powered agents can deliver measurable value. The objective is not to introduce AI simply because it is new. It is to help people work more efficiently, improve customer experience, strengthen operations, and support growth.

You can choose your speed and direction. You can begin with one process or several. You can move cautiously and still make meaningful progress. What is becoming increasingly difficult to defend is standing still.

Take a deep breath, organize the effort, and begin with a plan that can change as you learn. The destination may keep evolving, but the next responsible step should not remain a mystery.