Copilot Was Just the Beginning: What AI Agents Really Change

We Are Just Getting Started

Over the past year, Copilot has quickly become part of the Business Central conversation. It drafts content, summarizes data, and helps users move faster through everyday tasks. And to be fair, as a starting point, it’s pretty great.

But here’s the part I think we’re getting wrong: There’s this quiet assumption creeping in that with Copilot, we’ve “arrived.” That AI is now part of Business Central, box checked, and from here, it’s just incremental improvement. That’s not what’s happening. Copilot isn’t the destination; it’s just the beginning.

What’s coming next, and what we’re already starting to see, is something much bigger: AI agents.

From Helping Do the Work to Doing the Work

Copilot helps you do your job faster — that’s the value proposition. It responds when you ask, accelerates what you were already going to do, and removes friction. However, it still depends on you.

AI agents flip that model.

They don’t wait for instructions. They operate within defined boundaries and start executing. They make decisions. They move work forward in the background. That’s a very different relationship with your ERP.

The simplest way I’ve been explaining this is: Copilot helps you do the work, while agents start doing the work. And once you see it that way, it’s hard to unsee.

What This Looks Like in the Real World

This isn’t theoretical; we’re already seeing it show up in Business Central for many customers, and it’s pretty cool, TBH.

Consider accounts payable, for example. Historically, it has been a very human-heavy process: reviewing invoices, matching to purchase orders, and resolving discrepancies. Now imagine that process where the system handles the matching, flags only the exceptions, and quietly processes everything that falls within acceptable thresholds. That’s ‘doing the work’ for you.

The same pattern is starting to emerge in the supply chain. Instead of reporting on what happened, the system begins recommending, or even executing, replenishment decisions based on demand patterns and inventory levels. At that point, you’re not reacting anymore; the system is. Yeah, that’s awesome.

Finance is heading in the same direction. Reconciliations, variance explanations, and anomaly detection — these are all areas where the system is increasingly doing the heavy lifting, and humans step in where it actually matters.

Individually, none of these changes feel revolutionary. But when you step back and look at them together, it’s clear what’s happening: The system is taking on more operational responsibility.

Where Organizations Are Getting Tripped Up

This is where I think most organizations are a bit off track. The good news is that it’s early enough that you don’t have to panic, but it is time to get back on track and start moving the organization forward in a new way.

A lot of teams are looking at agents the same way they’ve looked at features in the past, something to evaluate, test, and maybe turn on later. But this isn’t just a feature. It’s a shift in how work happens. Treating it like a toggle, you can delay misses the point entirely. Especially if you are already super stressed and overworked, or that status is on the way…it’s time to get help.

There’s also a tendency to frame AI purely as a productivity story. Copilot fits that narrative well; it helps you move faster. But agents are about something different. They’re about capacity. They allow you to handle more volume, respond faster, and scale operations without simply adding more people or more effort.

And then there’s process. Haha.

If your processes are inconsistent, overly manual, or unclear, adding AI doesn’t fix them. It just makes the problems show up faster. A broken process with AI behind it is still a broken process, just running at a higher speed.

The Role of the User Is Changing

As agents take on more responsibility, the role of the Business Central user will start to change.

Less time spent doing transactions. More time spent reviewing outcomes, managing exceptions, and making decisions where judgment actually matters. That’s a better use of people, but it’s also a different skill set, and not one every organization is preparing for yet.

At some point, this stops being a technology conversation and becomes an operational one. Leaders must decide where the system can be trusted to act, where humans need to stay involved, and how much autonomy they’re actually comfortable with.

Where to Start (Without Overcomplicating It)

The good news is you don’t have to figure it all out at once. A practical starting point is to focus on one or two processes that are well-defined and relatively low risk. Accounts payable is usually a strong candidate.

From there, you can begin creating “safe zones” where the system is allowed to operate, and start shifting how you measure success, from volume and activity to exceptions and outcomes.

And just as importantly, you have to bring your team along. Because the move from doing the work to managing the work isn’t just a process change; it’s a mindset shift.

Final Thought

Copilot made Business Central easier to use. It is helping reduce friction and helping users move faster.

AI agents take that a step further. They don’t just improve how work gets done; they begin to take on the work itself. That’s a much bigger shift than most people realize.

We’re moving from systems that record work, to systems that help us do work, to systems that increasingly take on the work themselves.

If you’ve attended any of my talks lately, you’ve heard me say “Now the system is working for YOU, instead of YOU working for the system.” Welcome to the future, y’all, and we are just getting started.

---