Three Pillars For Managing Tomorrow’s AI Agent Ecosystem

I’m not a fan of expense reports. It’s an exaggeration to say they’re my mortal enemy in business. But I do see them as the epitome of a low-value, manual process that takes away precious time we could be using to accomplish more important tasks. 

I share my opinion of the wasted effort around expense reports in my latest book Digital Impact.  I also wrote that within two years, artificial intelligence (AI) agents will be doing the bulk of the work in filing them for us. We’ll give an AI agent receipts and other necessary information. Then, based on its training in our company’s policies, the agent will automatically complete, submit and likely even approve it.

Expense reports probably aren’t the worst in the grand scheme of things. But think of all the mundane, mind-numbing tasks your employees must do that get in the way of their actual job responsibilities. That’s precisely why you’re getting bombarded with so much talk about the “agentic” future, and vendors are rushing to release suites of AI agents.

These digital co-workers—autonomous software tools that can make decisions (based on their programming) to accomplish specific tasks—are already emerging in our businesses. In a very short time, they will be everywhere. The possibilities are limitless for using them to create a hyper-productive workforce and self-adapting processes, enabling smarter, faster decision-making while maximizing cost savings. We’ll leverage these AI helpers to remove tedious, time-consuming busywork from our plates. 

I envision a (very near) future where hundreds and even thousands of AI agents accomplish tasks done manually within our businesses today. Last fall, I predicted that agents would soon outnumber your human employees, and I’m feeling very good about that call. But this shift in how businesses operate also begs a question: How will you manage all of these agents?

Businesses are not yet emphasizing the long-term impact of these digital entities or how they will be governed. If not done thoughtfully, this will quickly become akin to herding cats. Not only is agent sprawl unavoidable, it’s already happening. As enterprises race to scale the proliferation of agents from multiple providers across operations, they’re beginning to see this new challenge. Businesses need a simple way to create, govern and orchestrate AI agents. And know how many are within the organization, what they’re doing, how they’re performing and whether they’re going rogue and hallucinating. If they misbehave, there needs to be a safe, fast way to shut them down without ill effects.

That’s why agents can become big problems if left unmanaged. The nightmare scenario is AI agents creating security gaps and data breaches, or taking unwanted actions. That’s why turning out agents without adequate processes for managing them leaves you bound to fail. With that in mind, here are three points to consider as you chart an agentic future that increases productivity without injecting chaos into your organization.

Agent home

You need a single, vendor-neutral space to manage all the different agents, from different providers, operating in your environment. Having them in one centralized place will give you the visibility needed to apply monitoring capabilities of their activities. Ideally, you should be able to interact with all of your agents using natural language, which will streamline collaboration and allow you to execute tasks in a safe environment.

Agent builder

If agents are to become ubiquitous in your business—and they will—you need an easy way to build and deploy them. You’ll need a design center that uses pre-built, no-code templates or can create agents from scratch with assistance from AI. With a simplified, prompt-driven toolset to manufacture agents grounded in your trusted data, agents can provide value faster and show a provable return on investment.

Agent command center

Every vendor is rushing proprietary agents out the door for customers. But how will you oversee all those different agents, as well as the ones you build yourself? You’ll require some kind of registry that provides visibility and control while proactively monitoring everything they do. Universal governance over your entire agentic force allows you to detect problems, discover anomalies, maintain compliance standards and mitigate security risks. Perhaps even more important, you can ensure agents from various vendors can collaborate on tasks across all business processes—not piecemeal on individual platforms.

Here’s the bottom line. Agents will need to be managed in the same way your business now (or at least should) provides oversight and optimization for your application portfolio, APIs and data. This kind of agent governance will ensure they’re acting reliably and responsibly, inspiring confidence and trust. I’m describing the need for your business to have an agent orchestration layer where you control these agents and ensure they work together.

AI agents are already here, but more are coming. Every week, there are announcements by technology giants and others about new, powerful agents. But your business won’t just have agents from, say, only Workday, or SAP, or Oracle, etc. You’ll have agents from all of them—and more. Managing them into a cohesive agent workforce will be the key to achieving your agentic aspirations. The people and organizations that embrace and use AI will stand apart.

It’s easy to become enamored with the possibilities of agents and envision how they can help make our businesses hyper-productive. But you have to think about managing them first. Or you won’t know how your new expense report agent is performing.