The Hidden Risk in Every Enterprise AI Vendor Contract - with John Belden of UpperEdge
The AI in Business Podcast · 18 May 2026
Source: https://podcast.emerj.com/the-hidden-risk-in-every-enterprise-ai-vendor-contract-with-john-belden-of-upperedge
Transcript: https://podcasts.opennash.com/shows/ai-in-business/the-hidden-risk-in-every-enterprise-ai-vendor-contract-with-john-belden-of-upper-157120/ (via OpenNash Podcasts, machine-generated)

Welcome everyone to the Emerj AI in Business podcast. Today's guest is John Belden, Chief of Research and Strategy at UpperEdge. UpperEdge is an independent advisory firm specializing in IT sourcing, contract negotiations, and complex project execution. In today's conversation, John discusses the six dimensions of uncertainty currently weighing on CIOs as they make large-scale AI commitments. He argues that the next five years of enterprise IT contracting will be defined by flexibility and adaptability rather than fixed productivity targets.

And he outlines specific contract mechanisms that buyers can put in place to preserve optionality and surface real progress. Today's episode is sponsored by UpperEdge. According to Nielsen, 91% of podcast listening happens alone, indicating deep, undistracted attention. Ideal for complex B2B messaging. To learn how leading brands and AI startups connect with enterprise AI buyer audiences at scale, download our media kit at go dot merge dot com slash partner. That's G o dot e m e r j dot com slash p a r t n e r.

Now, the conversation with John Belden. John, welcome to our Emerj AI in Business studio.

Yeah, I'm happy to be here, Yolandi.

Uh, I want to start with the mood on the executive floor right now. Um, for a couple of years, everyone was chasing the hype and greenlighting every pilot that they could find. And it feels like that ball is finally coming. Um, so I know that you're sitting with CIOs spending billions of dollars, uh, a year on programs that you're advising on. What are the real macro pressures pushing them to rethink the ecosystem they signed up for two or three years ago?

Yeah, I I'm gonna say there's, at least from my perspective, there's there's two very significant pressures that are sitting on the CIO right now, right? One of them I'm gonna refer to as the everybody else is doing it pressure. I mean, if you read the articles, right, if you read, uh, you listen to the your podcast, right? Everybody is talking about AI, everybody's talking about the, um, you know, the the potential benefits of AI. We've got, um, you know, all of the capital investments that are being made all over the world.

You know, boards are asking CIOs, What are you going to do about AI, right? So that's that, you know, that that big, what are you going to do with an AI? Question. The other pressure is is if I don't do something with AI, I'm gonna end up, you you know, in big trouble, right? And if you just think about it from the security standpoint of what's happening out there today, um, and you know, let's go back to Mythos and and Anthropic, right?

They're holding back that particular model right now because of the security concerns that it that it has in it. If you're a CIO and you're sitting in your seat saying, I'm just gonna wait and see what happens, you're a sitting duck, right? From the standpoint of what's gonna happen, um, when other companies or or bad players get a hold of AI and start to turn it against your company. So it's kinda you're damned if you do, and you're damned if you don't, right?

And and so I think most CIOs are sitting in that pressure seat right now of I have to do something.

That absolutely makes sense, and I think the pressure comes from, uh, it comes from the board, it comes from the C suites, it comes from, uh, just your peers as well in the uh, uh, in this environment. Um, so across all the cases that you've studied, what patterns consistently show up in in in this ecosystem?

Uh, patterns in in what respect?

So patterns in what we see, uh, them having to make decisions that they don't really know what the decision is about, or they don't know what to study before they make the decision. So they have these two questions that they're asking, but do they know where the information comes from to make the decision?

yeah, I think it's fair to say, you know, what I I think the the the biggest challenge, right? Okay, there's pressure that's sitting on top, right? But when we think about the the decisions that they made, the biggest challenges that they have, uh, I'm gonna put it, uh, in the categories of uncertainty, right? And if you go back ten years, right, you go back to ten years and you think about uncertainty ten years ago, um, there would have been perhaps uncertainty, uh, like a big ERP transformation, right?

There would have been uncertainty associated with, um, you know, well, what's what's this gonna actually cost me, right? There would have been uncertainty associated with, um, you know, does this software platform have exactly the capabilities that we need going forward? Fast forward now, right? Um, I'm gonna say you've got three pillars of uncertainty that didn't exist ten years ago when when people were actually getting ready to to do a big transformation. I'm gonna say the first one would be technology, just straight up technology uncertainty.

You know, if you could predict two years ago what models were gonna be capable of today, people probably would have thought you were, you know, it was wi witchcraft, right? There's no way that it's gonna look like it is today. Well, now CIOs are sitting here trying to say, well, what's technology going to look like two years from now, right? And who's gonna be the winners and who's gonna be the losers, right? Ten years ago, if you said you got to pick an ERP system, you would have said, well, okay, it's got to be SAP, or it's gonna be Oracle.

Those are the only two guys out there. If you sat here today and said, Well, who are you gonna pick for the orchestration layer? Well, are you gonna pick Google? Are you gonna pick Anthropic? You're gonna pick Open Ai, you're gonna pick Microsoft, you're gonna pick SAP, right? I mean, the the the platform uncertainty right now, uh, is gonna say is is is huge risk for the CIO. The second, let's call it the second layer of uncertainty.

Um, I'm actually gonna end up going with six instead of three, but let's let's go there, right? They the the second layer of uncertainty is environmental. And and what I mean by environmental, regulatory uncertainty being one, right? What are companies, what are governments going to allow and not allow? And then let's put on top of that, um, the potential for, uh, labor unions and how labor unions actually say, you know what, we're not gonna allow AI to come into these companies if you want to sign us an agreement.

And so therefore, you may be starting on a project that ultimately will be, uh, you know, negated by a labor agreement. Pricing uncertainty, right? When you used to be able to go in and buy, you know, buy an ERP package. Oh, by the way, I'm signing up for this license. I know exactly how much I'm gonna pay for this thing. Well, now you're sitting here watching the the prices of AI, while the per unit costs are going down, the utilization rates are going up, right?

And and so predicting what I'm gonna ultimately buy is, it's hard to do, right? Then you've got the next one, which would be implementation uncertainty, and let's talk about implementation uncertainty. These systems integrators today, right, with their implementation processes, they're embedding AI into the implementation processes. So how does that new implementation process work with AI embedded, and what's the governance structure that I need to put in place on that? And oh, by the way, what's it gonna cost me to support, right, a new AI implementation?

Because nobody understands necessarily what that is. Then you throw on, I'll call it the the the fifth one. What business model am I actually trying to shoot for, right? You know, what? What is the business case? Am I shoot? Is is AI gonna fundamentally restructure, right, my industry? And if so, what does that target look like? And then finally, you come back to the last uncertainty is do I even have the internal talent at my organization, right, to put on a project like this?

And what does that internal talent look like? So when you're sitting there as a CIO, right, you've got all this pressure coming down. I've got these six dimensions of uncertainty sitting there, right? How do I go forward and actually make good decisions in that environment? I mean, that to me encapsulates kind of the the pressure of the of the moment for the CIO.

I think it it gets scarier the more we we talk about it, because I think if we we have to think about the seven more, we might even identify a seventh or an or an eighth pillar there as well. And we're starting to see more high profile cases where massive digital transformations don't just miss the mark. Um, they actually sometimes cripple the company's ability to ship products or to to invoice customers. So there's definitely growing tension where service providers are using new automation, uh, and tools that work much faster.

But and I'm gonna get on that costing that you just mentioned. The clients aren't seeing those savings reflected in their fees. Uh, in fact, I think the contracts often feel like they're written for an entirely different era. Where are the hidden risks in these legacy agreements that this new wave of technolo- technology is actually making things worse?

Yeah, so so if we're if we're referring to the systems integrators, um, primarily, yeah, I mean, you're absolutely right. The contracts are still written old school, right? And and you know, by deliverables per se, right? And, uh, I'm gonna be honest with you, right? There isn't a perfect answer to your contract, your your, your contracting question right now, because of, I'll call it the the flexibility and what's actually evolving within the marketplace. Um, what we believe is important is more than anything creating contracts that incent flexibility, right, and incent adaptability to change as an organization actually learns, right, what the answer might be, right?

And so SIs are oftentimes incented to deliver what they actually proposed on, right, you know, six months ago, and we end up locking them in, right, to here's the deliverables, here's the timeline, here's the pricing for all of that. Fast forward three months, right? Fast forward three months, their processes has has changed, your target may have changed, right? The ability for you to actually do parts of the job that you couldn't do three months ago has changed, right?

So how do you craft an agreement with a systems integrator that is more focused not on holding the vendor accountable to what they proposed to six months ago, but more aligned with how do you go forward with a systems integrator? So that they are incented to bring the newest methodologies, bring forward the the latest risks, right, and adapt their process to a new direction that you might want to travel, which is a difficult thing from a contract standpoint, especially if you're thinking about it from a procurement where they want to hold them accountable, right?

I think in the next five years, it's not about accountability, it's about flexibility and making sure that you can adapt to how the market shifts.

so how can a C-suite person make sure about that? What is the question CIOs aren't asking their SI partners right now that they absolutely should be asking?

I would be asking the the CIO's number one, uh, the the SIs, I would be fundamentally straight up asking them for their roadmap of implementation of generative AI. We ask all the software companies what their roadmaps are for generative AI. I would be asking the SIs, what is their roadmap for generative AI? How do they think they are going to integrate it over their over the process? And how is that gonna affect your contract with them over the next two years, right?

You should be building in cost savings into the contract, you should be building in, I'll call it, uh, labor adjustments into the contract. Now, when I say built it into the contract, I'm gonna build that into the contract with a, um, I'll call it with a, with a caveat that says, hey, at least once every six months, we're gonna go look at the market and we're gonna see what the best practices are, and we're gonna make a, you know, we're gonna make an adjustment.

It used to be right that you could do a deal with an SI, and you could say we are going to allow an independent auditor to come in and audit your prices to make sure that those are still a a C, I'll call it, uh, competitive at the rate card level. I think the model's gonna shift that it's gonna be. No, we're gonna allow an an independent auditor to come in, and we're gonna audit you for your productivities and your capabilities every six months to make sure that those are still competitive, right?

That to me is is a part of the shift with how everything is actually going. I think it's entirely possible. And I and I'm gonna go back to my my old days when I was in manufacturing. We used to do deals with the big automobile companies that said, okay, we know that you're moving your um, uh, we know that you're moving your manufacturing offshore. And because we know you're moving your manufacturing offshore, our expectations are that our prices are going to go down year after year after year.

We're gonna put that into your contract as a supplier, right? You have to do that. Well, I think we should all be thinking about the same way with the SIs as they implement, uh, AI technologies. We know you are gonna become more productive, so therefore we're gonna expect efficiencies not related to repeatability, like in a maintenance agreement, but in implementation, where you're going to have different technologies to apply. Our expectation is is that we're gonna integrate them into our s into our process.

I like it. It sounds like we we're saying that we should tighten the accountability for, um, our SIs. Um, is that the the message that we're Yeah, t seeing?

tighten the accountability, though, around flexibility and adaptability, right? Not tighten the accountability associated with a particular, uh, a particular productivity number that says you are going to be this productive. Because I think I think that's a losing proposition for clients. Because I think if you go two years from now, that'll be an easy metric to hit, whatever that old productivity metric is, and all of the, you know, and all of the advantage goes to the S I.

I think what we have to do is anticipate that that productivity or those that adapt. I'm gonna call it, uh, I don't even want to call it productivity, right? I want to call that optionality, right? That there are going to be additional options that are going to be available to you. You want that S I exposing those options to you, where historically we've always turned that away and thought about it as, oh, he's just trying to sell me more stuff.

Well, over the next two years, it's probably to your advantage to listen to what those options are, because there's probably really, uh, I'll call it, uh, competitive edge that you might gain if you just listen. And I'm not necessarily a big advocate for the SIs. I just see the change that's happening within industries right now and how companies tend to get anchored on, oh, this is what we thought was the right answer two years ago, so we're still going down that road.

And I think that's a losing proposition.

And and it, it sounds like it starts at not just thinking about what solves the problem right now, but also looking at what will continuously solve the problem as things develop and and advance. I wanted to just also hammer a bit on that roadmap that you mentioned. You said that, you know, we should ask the SIs for their roadmap. Um, if a CIO does this and they get the roadmap, they get full transparency from the C from the SI, um, what are the red flags that they should be looking out for in that roadmap?

It's a great question. Let me frame, let me frame my answer a little bit differently. Okay, let me let me frame it in terms of questions you should be asking, okay, about what that roadmap is, rather than a red flag, um, in the roadmap. The first question I think everybody probably already talks about a little bit is there, you know, what is the protections associated with a human in the middle? Because there, there is always that that guardrail that says, uh, and everybody knows it, right?

I'm sure you've used ChatGPT, I'm sure you've used Claude, right? You know, if you know the subject, if you don't know the subject very well, you get a very authoritative answer. And it sounds like, well, I should believe it. If you know the subject really well, right? When you get an answer back, you can call B S on it because you know that answer, right? You know, so inevitably, you've got to have that that human in the middle right now.

And I would want to make sure that I understand what is the methodology that they are going to apply with the use of those tools. That's going to make sure that humans in the middle, I is in the middle, and I still have an individual that I can hold accountable for an answer at the S I, right? So that's kind of the first thing. The second thing I would be asking them about is I want to understand how we're going to introduce the context of my project into your tools.

And again, anybody that's used, you know, Claude, ChatGPT, understands how important context is to get the right answer. Well, when you're doing a very large ERP project, there's a lot of context associated with, you know, what's happening and and you know what this design looks like. And one of the phrases I like to refer to is how does your tool understand the context of now, right? At this point in the program, right, with all the decisions that have been made, with all of the factors that are in that are out there today, how does your tool take into consideration of the context of now?

And how is it not just saying, oh, you know, this was a a scenario three years ago that we applied it on, where that context doesn't exist. So, to me, that's probably the most, the two most important things I'm looking for, the human in the middle, and I'm looking for the context of now.

That makes sense, and I think the context of now is also something that we should revise with our SI. Probably, if you say we have to kind of review, um, and relook at the contract every six months, that's also a question that you should be asking every six months, not just at the start. Do you agree?

Yeah, exactly. You know, it's one of those things where, if you know, if you think about the way that these AI tools have worked, it has been dependent on the users bringing the context to the tool, right? You have to, you have to bring the context to the tool. If you think about it in the future, with maybe some of the things the SIs are going to be able to do, and and let's say the platform vendors, right?

In many cases, they're bringing the tool to the context, right? So so the context already exists, right now, how do I bolt on the tool to the context? And that, to me, that's the that that's the big, that's going to be the big thing that ultimately wins this orchestration layer, right? Is this context question, right? You know, the the the big providers like SAP and Oracle, they own the context today for the ERP system, right? They should be out there today, kind of talking about the context of now, and we've got the advantage of context of now, which Anthropic and Open AI and A WS can't compete with, at least that's my personal opinion.

And I I love seeing you get excited about this. I can see it's definitely something that you're passionate about. And um, before we get into the last part of our conversation, the there's a very clear pattern in your work where you like to distinguish between just trying to save a buck today and actually protecting a company's long-term flexibility. Now, flexibility sounds like it's it's kind of a big deal. Um, so right now, some leaders are just basically project managing their way through the next week, and some are just trying to keep their heads above water.

So we we want to dive into a an environment where we can do real framework building that govern these systems for the next decade. From your vantage point, what are the mental models that separate a leader who is actually in control of that transformation from one who's just reacting to the latest craze or the latest project delay?

You know, it was, uh, it's. It's interesting, Yolandi, because I, I, I did a. I've done a lot of research on failed transformations. Um, and and if I'm gonna take just a short aside to tell you a little story. One time I was standing in front of a, um, of a CEO, and we were talking about risks for big transformations. And I had just gone through this great big decomposition of risks. There's probably 900 risks on here, right?

And the CEO pushed his chair back and he looked at me and he said, John, he says, I want to know the one thing that I need to get right on this project, right? And my first thought was, Oh, this guy's an idiot, right? Because how I just told him there's 900 cars that can hit this guy in the road, and he wants to know which one he has to take off, right? So I probably made up some answer about, you know, you gotta get talent, right, or or something like that.

But I took away from that the the kind of the thing that says, you know what, that really deserves an answer, right? That really, really deserves an answer. And it's not, you know, it's not better execution, because you can go out there and find a whole bunch of companies that executed really, really, really well and still ran their company into the ground, right? And so and so I took it upon me to say, I want to study failure, I want to understand why companies fail, all right, and not what went wrong, but why it went wrong, right?

What was the what was the fundamental understanding, what was the fundamental things inside of the program itself that caused people to make the decisions that they did in terms of, uh, of of their program. What ultimately came out of it was there was, I'm gonna say, two things that that c that big projects really need to do. Number one is to understand and preserve their optionality, especially when we're talking about AI. We said there's so much uncertainty out there right now, right?

So companies really need to be able to say, how do I go through this project and keep my options open for as long as I possibly can, right? And if I do make a decision, how's how much time do I have to actually turn around and make another one? So I'll, I'm gonna say, keep my options open. The second thing that co that big projects need to do is they need to figure out a way to continuously learn on the pace, right?

That's best for them. Most companies, when they start on these big projects, they ultimately suppress learning, right? They they suppress the message that says, Oh, I don't want to see bad news, right? I mean, that you you end up with these biases that actually say, no, no, no, we can't tell the steering committee that. No, no, no, we can't say that, right? And it ends up with, I'm gonna call it performance theater, right? That says we're gonna stand up and we're gonna say that we're making great progress, and they suppress the learnings, right?

What you really need to be able to do is you have to be able to say, I've got all these uncertainties. I need to design my project so that I'm continuously trying to discover what the answer is of those uncertainties, so that I can actually change the course of my program before it's too late, right? So to me, it's not execution anymore, it's about optionality and learning.

That makes that makes a a lot of sense. Um, I think M. This conversation has been so insightful. I actually don't want it to end. We we are running out of time, but I think s some of the things that our executive audience will take away from this is. I loved our opening question where you said, basically, we have two questions or two pressures. Uh, one is what is everybody doing, um, at the moment, and the other is if there's no AI, I'm going to be in trouble.

So that's the first thing that we know. Okay, this is what we need to manage, and this is where we need to to start thinking about. Are we thinking about it? Uh, with sober mind, basically? Um, I liked in our in our second part of our conversation, how we discussed the idea of having a very flexible contract with our SIs and knowing to have to review that. Don't just it's it's this. We've seen the AI industry change, I wanna say, twenty times in the last two years.

So you cannot lock in vendors. It's it's just not it. It makes sense not to even think about it. And then, like you just said, keep your options open, keep on learning at a pace that is realistic for your organization and ultimately know your business. Is that is that the key message we wanna send today?

Well, know your business and know your industry, right? Know your industry because because the target that you're shooting at is probably gonna, uh, is probably gonna change, right? And that's one of those great big uncertainties that you say. I have to keep my finger on the pulse of what's happening out there. I do not want to design a product that is not gonna be useful three years from now.

And that is crucial at this time. Uh, John, a final thought that you would like to leave our audience with today.

I think this is the most exciting time that I've been involved with in the last thirty five, forty years of my career. I cannot wait to see what happens over the next five years.

I can see and feel and hear that excitement, John. Thank you so much for sharing that with us today. Um, I cannot wait to to get into this conversation some more in the future.

Very good.

Wrapping up today's episode, it's time for our three key takeaways from the conversation with John. First, the pressure on CIOs is no longer about whether they adopt AI, but about navigating six layered dimensions of uncertainty that simultaneously redefine what a successful transformation looks like. Second, the next era of enterprise IT contracting is built on flexibility and adaptability rather than fixed deliverables, with mechanisms like six-month capability reviews and independent productivity audits replacing the old rate card model of accountability.

And finally, the leaders who win the next decade of transformation will be the ones who deliberately preserve optionality and design their programs to surface bad news early, treating continuous learning as a structural feature rather than a soft skill. Position your brand alongside the Fortune 500 leaders defining the enterprise AI roadmap for the opportunity to showcase your solution to the executives currently funding and scaling global initiatives. Partner with Emerj. Secure your partnership at go dot Emerj dot com slash partner.

That's G o dot e m e r j dot com slash p a r t n e r. For further executive level analysis and to join our network of leaders delivering work for impact with AI, visit Emerj dot com. On behalf of the team at Emerj, we'll see you on the next episode.