AI Leadership Insights: Operationalizing AI

Mike Vizard: Hello and welcome to the latest edition of the Techstrong AI Leadership Series. I’m your host, Mike Vizard. Today we’re with Justin Mescher, who’s BP of Cloud Solutions for ePlus, and we’re talking about, well, what does it take to be ready for AI? Because a lot of organizations at least are kicking the tires, as they say, but it’s not clear to me that they’re entirely ready. Justin, welcome to show.

Justin Mescher: Thank you for having me.

Mike Vizard: So when you talk to customers, where are they on this journey? I think everybody probably is using some form of generated AI to craft a memo or maybe write a customer email, but that’s a long way from embedding it into an actual business process. What’s your sense of our relative maturity here?

Justin Mescher: Yeah, I’d say that the majority of clients I talk to are in what I call the AI-curious phase. They’ve got aspirations. They may not be well documented, they may not be honed in yet, but they’ve got aspirations of what they can do with their data to create business outcomes, and they’re trying to figure out what it takes to actually get there.

Mike Vizard: What is it that you guys are bringing to that conversation for folks? Because I get that they probably don’t have a lot of internal expertise, but when they look outside, there’s a lot of choices in terms of where you might want to lean on for some help, but what kind of help are they looking for first?

Justin Mescher: Yeah, a lot of times they’re looking for help just cutting through the noise. Everything’s AI these days. You look at any advertisement in any company, they’re all AI related. So how do you cut through the noise and identify what’s the right thing for you? So we always start from the envisioning side, think out of the possible. You’ve got access to all this data about your customers, about your products, about your business. What if you could do this?
So we bring a lot of industry-specific use cases in, help them envision what might be great for them, and then we start working backwards from that. If they identify an outcome that looks like it’s going to be beneficial to the business, we start identifying how far away they are from actually bringing that to life.

Mike Vizard: To that point, I think there was a fair amount of, shall we say, irrational exuberance last year around AI and every business leader in the world thought that somehow or other the magic was going to happen and we could have billion-dollar companies with two employees. I think a little more reality is setting in, how do you help prioritize projects for folks because at least 20 things somebody could think about doing, but they probably only have the time and resources to do maybe one or two, right?

Justin Mescher: Yeah. It really comes back to the business alignment on what are you trying to achieve. I compare AI a lot to cloud. A handful of years ago, people were putting things in the cloud just because they wanted to go to cloud, and they weren’t necessarily getting the outcomes they were looking for because they never really planned for an outcome. AI is the same way.
There’s a lot of people that want to “do AI right now,” and you ask, what is it you’re trying to accomplish? It’s just proving something out, but they don’t necessarily have an end in mind. So a big part of it right now is just identifying that first proof point. You’re not trying to eat the whole elephant at once. What’s something that you can prove out? Identify, “Yes, we can create business value that’s going to help me go get additional budget, additional headcount, additional access to technology.” So prove it out, show that it is going to be beneficial, and then go after other bite-sized chunks.

Mike Vizard: Is there a divide between the data scientist community and the business folks? I mean, there’s a running joke about how the data science team created this awesome model that showed that revenue declines every seven days and the business people told them, “That’s because we’re closed on Sundays.” So how do we get those data science folks to understand the business that they’re trying to augment?

Justin Mescher: I would say rather than divide, I’d call it a disconnect a lot of times. I think that a lot of times comes from you’re going to have your really smart technologists sometimes have a disconnect from your really smart business people. They don’t necessarily speak the same language. They don’t necessarily have the same care-abouts, and the secret sauce is getting those two things to connect because data scientists love to solve problems, love to solve complex problems. Business people have complex problems.
Identifying how to take one of those complex problems from the business and get it to a data scientist, make that the challenge they want to take on, give them very specific metrics you want to bring back that helps bring back the data you’re looking for. To your point with that joke, sometimes you’re going to bring back data that without context doesn’t make sense. So it can’t be a one-way, “Here’s your instructions bringing me back the details,” because without that additional context of, “Oh, yeah, we don’t work Sundays, so let’s not look at why one seventh of our time isn’t being used properly. Let’s iterate and let’s work on it together,” and really creating that connection between the data science team and the business is really key.

Mike Vizard: What do you expect for the coming year? My sense of things as we go along is we’re going to spend the year trying to learn how to operationalize AI and maybe the ROI might not show up until 2025.

Justin Mescher: Yeah, I think this is the year of the POC and the year of budget allocation for a lot of companies. Don’t get me wrong, there’s a lot of people that are doing really big things with AI right now, but if I were to look at the everyday organization that’s just in the curious phase, I think right now is about making sure you got your envisioning, you got your business-aligned strategy, and that you’ve got some sort of proof point that you can go out and try.
The other trend is Microsoft 365 Copilot. That has been something that has created a ton of chatter because when you log into your Office applications, it’s just sitting there in the corner saying, “Hey, click me. The magic’s going to happen.” I think because that’s more consumer available, it’s to drive just a lot of awareness inside of businesses that, “Hey, this is something I need to look at even if AI isn’t some sort of massive business-changing thing, analyzing all my data, maybe it’s making an end user’s life more productive from day to day. Maybe that’s my proof point.” So I think that will be one of the primary entry points we see.

Mike Vizard: It seems to me there’s going to be a copilot or a digital assistant for everything. How are we going to stitch together all these different assistants and copilots into something that feels like a workflow? I mean, ultimately, I think what we want is for my copilot to call your copilot and have something happen. Right?

Justin Mescher: Absolutely. You know what? The one thing it starts at is data. Because first thing, you need to make sure that your copilot is working the right way before you try to get it to go out and work with other copilots. The other consistent theme we’ve seen with clients right now is if you look at a hypothetical example of how do you feel about your data governance today, and if you were to expose your data to a copilot type of technology, do you feel that it would only expose and bring back data that that person should see? The answer to that is usually, well, not really.
So I think first is getting your copilot turned on the right way and making sure that things are properly governed. It is the most efficient it can be. Then you start to federate out into other parts of your organization. You feel comfortable with that governance. Now you can federate potentially into other third-party organizations, but I think conceptually it’s get your house cleaned first and then look at how you can federate out. Any of this stuff turned on too quickly or federated too quickly is going to create exponential risk beyond what we’ve seen in the past.

Mike Vizard: All right. I think you just said those of us who live in AI glass houses should not be casting stones, but I get the idea. How does these all come to fruition in this regard? I think this is a little far out, but I’ll pose it anyway. So let’s just say I had some sort of AI model that is optimized for a process. Let’s say, I’m the buyer and I’m trying to get the best deal, and I’ve optimized my AI for that, and my model encounters your AI model as a seller and you’re trying to get the best profitability for your thing. Won’t these two AI models just beat each other to death and then we’ll have to call a human and intervene? How is this all going to play out?

Justin Mescher: Yeah, I think you hit it with the last part of the comment there is AI is there to be an enabler for human beings. I think those are looking at AI as an absolute replacement. I think that’s so far future aspirational. It can replace certain human functions, absolutely. But without oversight, without context, without emotion, without logic, without reason, there’s not a proof point that that is something that can live on its own.
So I think one of the biggest trends that I’ve seen in organizations that are adopting AI quickly is they’re doing it in a very responsible fashion. They’re doing it in a way where humans are governing it, monitoring it, influencing it, instead of trying to turn over the keys to the kingdom. Because the example you gave there is a perfect one in that those standoffs will start to exist very quickly.

Mike Vizard: What’s your best advice to folks then, or conversely, what’s that one thing you see folks doing today that just makes you shake your head and go, “Folks, we need to be a little smarter than this”?

Justin Mescher: Yeah, I mean, depending on what industry trade rag you want to read, 60, 70, 80% of AI initiatives are failing right now due to data issues. I think number one, I said it earlier, it’s just get your house clean. If you’re going to start leveraging AI and exposing your data to it, you need to break down data silos. So you need to make sure that you’ve got some sort of modern data platform that is prepared for this.
You need to get your governance in order, ensure that you’ve got people and processes only have access to the data sets they should, because your blast radius on risk becomes much larger when you start to turn just autonomous things against it. So it really is get your house in order first and then take a small step and then take another small step, I would say is my number one advice right now.

Mike Vizard: How do you have that conversation? Because it’s awkward, and here’s how it goes. It’s right. So the company says, “We ought to do this AI thing,” and the IT guy says, “You’re right, and we’re going to need better quality data.” Then the business side, “But I thought we had better quality data, and I thought that’s what you were doing for the last 15 years.” But then it turns out that we lift up the rug and the floorboards are rotten and the data’s a mess. So do we need to have maybe a global forgiveness moment for data management? But how do we get there?

Justin Mescher: I think the reality is we’re all technologists and technologists rely heavily on humans. That is that there’s trust factors in there. You got your HR handbooks on things you shouldn’t do. You’ve got all these legal ramifications. You got these things that are put in place where even if data governance wasn’t perfect, there was a trust factor, there was a policy factor. There’s those things that we put in place that made it okay, made it acceptable.
When you remove that human factor, you potentially start to remove ethic-checking and things like that that you can’t govern quite so well out of the box. Now all of a sudden, you really have to make sure that the cleanliness is got to be spick and span. So I think it’s less about the fact, “Hey, we didn’t do our jobs previously.” It’s more, “Our jobs changed, our responsibility to the business changed, and then we have to have absolute cleanliness, clarity, visibility into these things.” Whereas before, a 90% clarity on that might’ve been okay.

Mike Vizard: What is your sense of what’s going on with the regulatory folks? Because a lot of business people are like, on the one hand, they’re like, “Wow, full speed ahead and we can do all this wonderful stuff.” Then they have this little feeling in the back of their neck. They go, “Hey, is all this stuff we didn’t about to be legislated out of existence? Because there’s a regulatory body out there that hasn’t quite put the rules in place yet.” So do we need the rules to come faster and are the rules going to be reasonable?

Justin Mescher: I think the rules are going to come over time. I mean, there’s still rules being developed around cloud, that cloud’s been prevalent for a very long time now, and there’s still a lot of rules and regulations catching up, and there’s things you have to do. Just be ready to be agile when those new things come down to make sure that you can react as you need to.
The intellectual property is a great one right now of when you are using Gen AI to create new content, where’s that intellectual property live? There’s plenty of lawyers and plenty of regulatory bodies working on those things right now. So it’s all about making sure your eye’s wide open and ready for what’s next. There are going to be a lot of pivot points, but I think it’s not just something you’re going to see in 2024 here. It’s something you’re going to see for the next three, five, seven years.

Mike Vizard: All right, folks. Well, you heard it here. We’re all on the same road. We’re just not quite clear exactly where that road goes, but we’re going to find out together. Justin, thanks for being on the show.

Justin Mescher: Excellent. Thanks for having me. Appreciate it.

Mike Vizard: Thank you all for watching the latest episode of the Techstrong AI Series. You can find this episode and others on our website. We invite you to check them all out. Until then, we’ll see you next time.