(This is an adapted excerpt from Boomi CEO Steve Lucas’ book, “Digital Impact: The Human Element of AI-Driven Transformation.”

When the Apple App Store opened in 2008, it had 500 applications you could download on your iPhone. As of 2024, the number was approaching two million. There are more than 2.6 million apps available for Android devices. I predict that the coming explosion of agents won’t be like what we observed with mobile device apps. The growth will far exceed it. There will be an Agent Economy.

This will inevitably lead to agent sprawl.

Your organization will have another business object within the architecture to manage and orchestrate, just like applications, databases, and APIs. And you will be accountable for the actions of these agents as part of your overall IT architecture. This is why businesses need to know what these agents are, who installed them, where they live, what they decide, their training methods, and whether they’re hallucinating. Are they meeting governance requirements? What about standards around personally identifiable information? In other words, who is running the agent asylum?

With great power comes great responsibility. (Alert readers will recognize that I paraphrased the proverb Uncle Ben popularized in the Spider Man movies when he gave that advice to Peter Parker.) This line bears repeating when we think about agents. They will be incredibly powerful, and it will be your responsibility to ensure that the power is helping your business and not hindering (or endangering) it.

Steve Lucas

Leaders need to know everything about agents because they are essentially digital employees that work for your company and represent you in the real world. Without management, mischief is possible, even if it’s completely unintended. You’ll need a fundamental understanding of everything about them.

Public marketplaces exist where anyone can find and download agents to accomplish specific tasks. If your organization is advanced in AI, you might already be building agents internally. I suspect those earlier adopters can figure out where I’m going here. You need a structured system to manage agents within your architecture.

Agent Registry

If you think of agent marketplaces as a version of the app store where you download the agents, the agent registry is your iPhone. It’s the place in your architecture for containing agents. It’s where all things agents happen and where you can keep tabs on them.

You will need this place to track and manage their activities. You will need an agent dashboard, activity logs, and runtime statuses. You will require a registry of every agent and comprehensive reports on what they’re doing to ensure they’re not running amok. It comes down to agent explainability. If these autonomous and semi-autonomous agents are tasked to do stuff, you need visibility into precisely how they do it.

I look at it from a CEO’s perspective. I want to know what these agents have been approving, whether it’s expense reports or something more significant. I’ll use a hiring example that resonates with me I used to head a talent acquisition software company. Let’s say my business has an agent helping us identify top job candidates based on their resumes. I need to know the selection criteria and ensure they match my company’s values. But we can’t do that without transparency into how that agent was trained and how it came to decisions. What, for instance, if it began to select candidates for interviews based on, say, a zip code preference? Talk about the opposite of diversity, equity, and inclusion! This isn’t acceptable at any level.

You have to know everything about these agents, including if any biases somehow affect their choices. You want to be aware of every potential risk. So, it’s not a stretch to think that monitoring the activities of agents will be mission-critical for your organization.

The Agentic Future

Every company on the planet is busy creating AI co-pilot experiences that you can talk to and help you do some things. That’s all well and good. But it can’t be just something that occasionally does stuff when asked. The power of agents is that they’re always on, always working – without constant, hands-on management.

Now, there might be a sense of dread in the back of your mind when you think about the expansion of agents. What will this mean for jobs? My personal belief is that this doesn’t mean jobs will go away. I think AI and agents will augment people, not replace them. For me, the formula is this:

Humans + Machines = Greater Success

You might be familiar with the acronym HITL – which stands for “humans in the loop.” It describes the critical role of people in providing oversight and context for AI.

We are better together. In our personal lives, everyone will have some kind of personal AI assistant – i.e., an agent. They’ll be personalized to know who you are and remember your likes and preferences. They’ll know enough about us and our jobs to remind us each morning: These are the three tasks you need to do today at work. From a business perspective, agents will improve over time, making our companies more efficient.

Within the context of your business, and specifically around integration, automation, data management, agents will be invented and introduced to you in the next few years to perform tasks bordering on magic. Cleaning your data autonomously, data discovery, and more.

However, agents will also increase your integration needs by orders of magnitude as they require connectivity with critical systems within your digital architecture. This is why you need an AI-driven integration and automation platform designed for your business to create interconnected and automated processes for applications, databases, APIs, and agents.

Adapted excerpt with permission from the publisher, Wiley, from Digital Impact: The Human Element of AI-Driven Transformation by Steve Lucas. Copyright © 2025 by Boomi, LP. All rights reserved. This book is available wherever books and eBooks are sold.