Six Challenges to Scaling Agentic AI in the Enterprise

Ray Rike (0:08)

Welcome back to another episode of the AI to ROI Big Story. I'm Ray Rike, founder and CEO of Benchmarkit.

Peter Buchanan (0:17)

And I'm Peter Buchanan, managing partner of NewPlan. And this week, Ray, we're covering our September 15th AI to ROI newsletter story, the six challenges of agentic AI and how to overcome them. It's already been very popular on Substack this morning. So hopefully, this podcast will also be popular.

Ray Rike (0:39)

I'll tell you, it's a very common conversation I'm hearing in almost every company I talk to, and that is, how do we deploy and ensure we can get some true return on investment from agentic AI investments. But before we dive into all the details, we have a lot, let's hear a word from our sponsor.

This episode of the AI to ROI Podcast is brought to you by benchmarkit.ai.

If you're trying to figure out how AI is actually performing inside companies, the real numbers without the hype, Benchmarkit is the resource. They're the industry's leading benchmarking research firm for how companies are deploying AI and measuring the business value it creates. Their research covers AI maturity and best practices across finance, sales, marketing, customer success, operations, and much more. So you can see exactly where your organization stacks up against peers who are further along. And here's the best part. Benchmarkit's research reports and interactive benchmarks are all free at benchmarkit.ai. And while you're there, grab the big book of AI Metrics. It's a free guide built around a five-layer AI business measurement framework that includes 81 AI metrics across 13 functions. If you need a common language for measuring AI to ROI, this is it. That's benchmarkit.ai. Free benchmarks, free frameworks, free content, real data. Now back to the show.

Peter Buchanan (2:14)

We're back and we're going to talk about the first of these challenges, which is the production readiness gap. Let's set the stage there, Ray. Eighty-five percent of enterprises are now working on agentic AI use cases, but only five percent have put one into production that delivers meaningful ROI. It's like a ghost of the MIT report from last year.

That gap is the real story of agentic AI in 2026 It has nothing to do with model quality and only maybe a little bit to do with the agentic technologies that enterprises have picked. It's a pattern that shows up across every industry we track. It's worth diving deep on these six challenges of agentic AI and how to overcome them.

Ray Rike (3:06)

Hey, Peter, can I ask this? 85% are working on it, but only 5% have put one in the production. It delivers meaningful ROI.

Peter Buchanan (3:15)

Yeah.

Ray Rike (3:16)

It does mirror what MIT said. What specific research report did that come from?

Peter Buchanan (3:22)

I think that's from a Deloitte report.

It's sourced in the newsletter, for sure. I'll have the newsletter issue open right now. What I'd say there is there's a little bit of a caveat and that most of these agentic AI use cases that are at 85 percent here, they're maybe not done with them yet, which is why they're not deployed or they're still working out the kinks. I think enterprises get a little bit of a pass here, but it's still a big delta. Yeah.

Ray Rike (3:49)

We're definitely in the early innings of agentic AI and a couple of the mistakes. I know we're going to talk about six challenges, but why there's only five or maybe it's more than five percent of agentic AI initiatives actually don't have, compare really compelling ROI is number one, they don't change with redesigning the process to really take advantage of agentic AI and they don't identify the measurements that ultimately are going to show the ROI. So I think that's kind of where it starts. Now, I know even like at Cisco where they built this personal AI agent called MyAgent, and it's available for all 90,000 employees, and it's backed by like over, I think it's like 800 specialized agents.

They say that it's delivering real value. It's not a technology issue, but it's truly a measurement issue. So I think one of the things that companies really need to understand is what's the perceived and potential value right up front.

Peter Buchanan (4:53)

Right. So let's get to this production readiness. So Gartner expects the average Fortune 500 companies to be running more than 150,000 agents in production within two years. So basically the end of next year, or maybe the beginning of 2028 So you scale that fast, you create mistakes along the way. So in conjunction with that number, Gartner data says only 13% of enterprises believe they have adequate governance to manage those 150,000 agents. So that's not readiness at all.