XBP Global CEO Andrej Jonovic On Why Exception Management Is The New Enterprise AI Battleground

Enterprise automation has historically worked by standardizing what is predictable, then relying on experts when something new appears, and a judgment call has to be made. Andrej Jonovic, CEO of XBP Global, sees that balance changing as AI workflows become more capable and more adaptable. 

In this conversation, he discusses why exception management is becoming a central AI challenge, how companies can turn process knowledge into flexible workflows, and where human judgment still has to stay close to the work.

Why do you see exception management as such an important AI topic for enterprise leaders?

Much of enterprise automation was built around the same pattern. The system handled the standard cases, and people handled the rest. Over time, the organization became valuable because it understood the process, including the edge cases, the calls that had to be made when the system couldn’t resolve something on its own.

AI changes the economics of that. Clients are looking at the same work and asking a fair question: why can’t more of this be handled directly? In healthcare, for example, payers are investing heavily precisely so they can pull more of that workflow in-house.

So the real question becomes how you stay relevant when the client can do more on their own. If your value was built around managing exceptions, you’d better think hard about what that value becomes once the number of exceptions starts shrinking.

What changes when companies move from rule-based systems into AI workflows?

The client gets something more flexible than another layer of fixed logic.

Instead of hardwiring more rules into the existing system, you give the organization a workflow it can actually adjust as its needs change. They can add components. They can add services. They can keep improving the process instead of waiting around for the next rebuild.

Most enterprise problems aren’t new. The U.S. healthcare system hasn’t suddenly gotten simpler. What’s changed is how much of that work can now move with fewer handoffs, fewer touches, and a lot less dependence on the old rules-based structure.

A good AI workflow should cut effort and hand the organization more control over how the process evolves. And it has to be secure and cost-effective enough to actually run inside the environment the client has, not the one you wish they had.

Where do humans belong as more of this work becomes automated?

Humans should move closer to the exceptions that remain. That’s the short answer.

AI can prepare the work, analyze the information, and help push a process forward. But it still needs context around it. Someone has to understand what’s actually different here, what the client is really asking for, and what decision needs to be made.

That makes the human role more focused, not less important. Less routine work, less time spent catching up on information the system already prepared. The value shifts toward judgment, relationships, context, and the ability to know when the system’s answer just doesn’t fit the situation.

People get more valuable in that role, but only if leaders actually move them there instead of leaving them buried in manual work.

How should leaders begin identifying where AI can reduce exception work?

Start with the workflow, then choose the model around what the workflow actually needs.

Look at where people are spending time because the system can’t handle a certain case. Look at where fixed rules keep breaking down. Look at where your experts are making the same call over and over.

Then ask what has to be true for that workflow to run safely. What data can you actually use? Where does the model run? What standards does the organization have to meet? What’s the cost base you can live with?

That matters even more in regulated environments. AI is a lot easier to talk about in the abstract than it is to actually put inside a real operating model.

What does this mean for enterprise providers whose clients are trying to automate more themselves?

It means providers have to be honest about where their value is going.

If your value was built on knowing the process and managing exceptions, you can’t assume that model holds forever. Clients will keep automating more. Some will bring more of the workflow in-house. That’s already happening, and we see it.

The way to stay relevant is to turn your process knowledge into something the client can keep using themselves. Give them a workflow they can adjust, extend, and trust. Help them cut routine effort while keeping their people focused on the decisions that still need real experience.

That’s the shift: from managing the mess to helping the client make the work itself smarter.