Implementing Microsoft Copilots with Unisys’s Joel Raper

Hello and welcome to the latest edition of the Techstrong AI video series. I'm your host, Mike Vara. Today we're with Joe Raper, senior Vice President and General Manager for Digital Workplace Solutions for World Unisys.

And we're gonna be talking about, well, how to get started with ai. I think we've all been impressed and generally see the value, but boy, it's hard to figure out where, where to put your arms around this thing. Hey Joe, welcome to the show.

Thank you very much. I'm excited to be here. What do you see in folks doing initially with Gen AI and these types of tools?

Because I mean, beyond the fact that I can write a better email or maybe create a better document that's a little faster, um, how do I operationalize this stuff? Yeah, it's, it's a good question and I think everybody's kind of struggling with the use cases and the value proposition that comes with it. We can all see it in front of us.

We can definitely see the potential that's there, but the real world application, I think is Laing a little bit behind on the hype factor that we've got going right now. I think within DWS specifically, or digital workplace services, I do think that there's several areas that have a real and very near term, uh, opportunity for gen AI specifically, um, to take advantage of some things that will get customers or get, um, organizations set apart from, uh, taking advantage without the risk that comes with a security governance and risk. What comes with all of the other aspects, you know, one would be taking advantage of Microsoft copilot.

You know, Microsoft has spent a lot of money in open AI and the investment and then has rolled out copilot many, many organizations. Um, and we can see some fruits. I think you mentioned a little bit about creating a, crafting a better email, um, coming up with some images, maybe helping with the PowerPoint, but there's a whole lot of other things that can happen out of that.

And I'll give you a really good use case example that we have. Um, you know, if you're a project manager, you send out a lot of status reports, you keep task lists, you do a whole bunch of things like that. Well, right now, if you turn on copilot for example, um, it can send out the status reports for you.

It can keep your task list. You can add it to a, a team's client for example, and record all the to-dos and the major actions as well as the, the highlights of a particular video and apply that to your daily use. I think the challenge that comes specifically with that is how, what do you do with that captured time, right?

If you saved an hour a week, if you saved two hours a week, where is that usefulness of that time that offsets the cost of the creation of Gen AI and, and the data governance and all the other pieces that come with it? And I think that's the part most organizations are struggling with. And I think we have a couple ideas within Unisys on how to take advantage of that.

Um, and I'll share that more as we go on, but I'll, I'll pause there. I think part of the challenge is simply that there's a lot of things that we do as humans by rote. We just do it.

'cause we know that's what the job, we know what the tasks are and we know what they're organized. And for me to offload that to somebody else or an AI assistant, I actually have to stop and think about what it is I'm doing. And a lot of folks, uh, don't really have an ability to kind of, or stop, or the inclination to do that is, uh, it's not natural to them per se.

So how do I discover what my processes are that I might wanna offload to an AI assistant? How do I kinda wrap my head around that? Yeah, it's, it, that's probably the number one problem that when I talk to CIOs of organizations that they are, they're struggling with right now is that concept, right?

How do we take creatures of habit and embed this into their daily activities? And, and I'll give you a scenario that in our particular offering of rolling out, we'll use copilot, like as an example, you know, we separated two classes of users. So a, a first class of users was a group that we sent a bunch of articles to, here's some tips, here's how to use it and enable them for copilot.

Um, and then the second class of, of, of, of folks that we set up for this trial, um, we curated it, we talked to them once or twice a week. We sent them, um, ideas based on their persona or the type of work that they do, and said, these are the top three or four things that people are getting advantage of. We are, we also, um, monitor in the sense of what is the use.

So if you and I get really active in it and we take advantage of it for the first week or two and the third week it starts slowing down a little bit, we monitor that and actually said, Hey, we noticed you're not using it as much. Can we help you? Can we show you some tips and tricks?

Can we show you some of the new things that are coming out? And I think that kind of organizational change management helps a lot in how we change our creature of habit mentality and our consistent part of our job. Do you think we need some sort of workflow discovery tool and is that an AI tool or is that a process minor tool to kind of figure out what are our processors?

Because I'll be honest, you know, if you ask me about stuff, I'll say, yeah, we totally understand that process. And then the more I think about it, the more exceptions there are probably than there are rules. Yeah, I'm, I'm gonna date myself a little bit and talk about, you know, I think in the early two thousands it was really popular to employ six Sigma folks and people that were certified in process and automation improvement and everything else.

I think that that comes back even more so now than it, you know, kind of weaned off just a little bit to your point. Exactly. I think that we do have to look at what is the daily life, what are the activities that each persona or each job type does and, and try to come up with some AI advantages of them.

But more importantly to that is actually show the value creation, right? There's a lot of work that goes into your investment in learning your investment and all the things that comes outta it as an applicator of, of gen ai. But what's the value creation for that investment, right?

If I spend 10, 15 hours kind of training and learning, but it saves me eight there, that doesn't work for anybody. Um, just the only thing it works for is checking out the box that we're using gen ai. We have to actually show significant savings in it.

And so we're employing people just like that, that are, you know, six Sigma like or six Sigma certified people to look at each persona job type and find opportunities for us to employ a gen AI solution or, or even an ai. We, we lump it all, we lump it all up together, but there might be some machine learning, there may be normal AI as well as gen AI to impact this. I also struggle with the notion of I seem to talk to everybody and everybody's got their own AI assistant now.

So, uh, will there be one kind of master AI assistant from, I don't know, Microsoft or somebody else, and then a lot of, um, sub assistants that are trained for specific tasks. How, how will all this get orchestrated? Yeah, I, I see, you know, it's hard to tell what happens five years from now, but I think, you know, from now to to the next five years there's gonna be many, many, um, gen AI assistance or gen AI applications.

And some of those will come from, you know, like if you use Salesforce right now, they have a whole bunch of offerings around gen ai. If you use ServiceNow, um, in the ITSM world, they have a whole bunch of offering around their own version of Gen ai. And so I think there's gonna still be a lot of that, but I'll, I'll give you an example right there.

We use Microsoft copilot as I talked about already for office apps, but there's also kind of a studio copilot you can put in front of a SharePoint application and you know, that'll instantly give you access that you never had. Or the, the search function for SharePoint never worked very well. And that's a separate ai.

So I think we're gonna be in this world where, you know, we, we take advantage of many different AI applications or applications to ai, um, and, and whatever aspect of our jobs are, What is that workforce gonna look like going forward? It seems to me it's gonna be a mix of humans and machines. And the machines might have personalities, they might have avatars, they might even tell you a joke every now and again.

But is their workforce gonna be made up of, you know, people and machines with machines with identities? Yeah, this is the, uh, the, uh, the question of the day for sure. And everybody's worried about the impact of this.

Um, but I'll, I'll go all the way back to when computers came about, right? When you started throwing computers and databases and a whole bunch of financial systems in the case, like that was gonna make everybody's job easier, right? It was gonna automate so much of it.

Well, what did it create an entire industry of, you know, network management and IT people and everything else? I think we're gonna have that same concept. It's hard to see the, the crystal ball ball on what that looks like, but I think, you know, the world is going to be, um, in a increasingly more complex situation and require a lot of people to take advantage of that, to train, to take care, to build, to do the, still the same things that we had prior to, um, this big push for ai.

I think, um, you know, in the near term you, you're gonna have an entire generation, uh, not generation I would say, but a workforce that is, uh, very skilled in the prompting in the, um, the, the figuring out what the capabilities of AI and trying to bring that into what is your normal process in a day. I think that's gonna be a big boon for the industry in the short term. But I think as it goes, it'll be just like everything else, the more complication that comes on, yes, we can accomplish more, but we'll get more to work on and we'll have to use more and more systems to achieve our day job.

Yeah. And there's some obvious concerns about, you know, being automated out of a job and all that other things. But I was looking at some math also that suggested that there's simply aren't gonna be enough people to drive all the tasks that we need to get done to keep the economies growing at the rate that we're growing at.

So if we wanted to maintain a 2% growth rate in the US for example, we would need like another million people to show up for work. And last time I checked they weren't coming. So, you know, are we kind of, we're at a point now where we've just run out of bodies.

Yeah, I think there's another philosophical question, right? On one side you get people like Elon Musk talking about, you know, we're, we're not having as many children and there's gonna be a population decline and all of that. So that's the opposite side of what you talked about.

And I think in, in the other side of it, I think, um, there are some workforce challenges, but I think gen ai, um, and the focus within the enterprises themselves are gonna kind of change and transform of what people do. So if I, you, your, your point is very spot on. If I've got a 40 hour work week doing exactly what I do right now and gen AI will save 10 or 15 hours, I, I made a fairly impactful, uh, um, impactful time savings in there so I can do more things that comes along with it.

So I, you know, who knows what it's, it's actually gonna end up. But I think there's somewhere some mixture of it. It's always fun to talk about this stuff.

Um, but whether it's, you know, we don't have enough population or, uh, we have too, too much either one, I don't know. 'cause you know, there's another big talk to that's the opposite side of what you're talking about where we're gonna have to create a universal income because of gen AI taking all of the work. So who's right?

Hard to tell. I think time will tell on that one. How or jobs evolve?

'cause it seems to me right now we hire a bunch of people to perform a task and as this situation continues to unfold, it almost seems like we're all gonna evolve into supervisors of tasks that are handled by machines to a certain degree. And we will manage the orchestration and make sure that the things are done as anticipated, but maybe we won't all be doing as much. You know, even your average office worker has a lot of manual step work that they gotta do every day that maybe we won't be doing.

Yeah, I mean, I think that there's, there's a world of, um, of physical work, right, that comes out to it. We have a very large, uh, field services organization that goes out and does break fix repairs and, you know, if you're fan and your laptop dies, it's us doing the service. It may be both doing maintenance tasks and repair of other devices in the workplace.

You know, until we get robots, that's not gonna change on that side of it. Um, but your point is I think pretty valid on, you know, are we just gonna be supervisors? I I heard a story from a very large Fortune 500 company talking about the majority of all their contracts now, um, go through the creation of a contract through Gen ai.

And you know, this is, I think the, the, um, things that we as humans need to think about if you do that, who is going to cut their teeth to learn how to create contracts and to know what good looks like and what bad looks like. We, we read in the industry, and if you've used, you know, some of the tools that are available right now, gen AI doesn't give you a hundred percent of a correct answer, right? And so there's a lot of things that can be portrayed incorrectly or the data's incorrect or anything that comes out out from it.

I think we need to be very conscious of not losing the, uh, the entry level skillset that for that person because, or for that profession, because then we won't have the experts at the end to supervise it. And, and so I think this is a really, really thing that we all need to keep in the back of our head, that it should help us make decisions. It should take away rudimentary task.

It it should provide us with more data or insights for us to make decision, but we actually have to go through the steps to have that capacity and knowledge and experience to make that decision. And, and I think that's the little bit of the catch 22 we need to be worried about right now. And there's a subtle thing about all these AI assistance.

One, they're probabilistic, not deterministic, which means that they will probably give you the right answer and they will probably say that with a great degree of confidence that makes you think that they know what they're talking about. But in reality, there's gonna be instances where whatever they're recommending isn't the right choice of the day. Uh, they may get smarter, but ultimately we have a lot of processes that are deterministic and that they're gonna be right a hundred percent of the time.

So how do we kind of strike that balance? Do we need people to have just a better sense of AI literacy to understand that? Or is there some way to get after that, a little bit more nuanced approach to managing these workflows?

Yeah, I, I think there's, there's two flavors, right? We're used to, that's very evident. If you use chat GDP or any of the, the technologies out there, right?

You get whatever it's been ingested by the internet, right or wrong. It's not like all of it came from the encyclopedia. Um, some of it came from Wiki for example, right?

And we know that we can, um, you know, manipulate what Wiki says about certain topics. The other side of it though, and, and gen AI specifically as it relates to maybe ITSM or the ticket information of a service desk or the information about your assets that you have in an environment, um, that will be more right, because the data is, is curated, it's governed, it's controlled, and it's very specific to these things. And so I think that we'll have more success when we think about those aspects of it than we take a general AI with a massive large language model and a massive index like the internet, um, in that scenario.

So I think you got two sides of that fence. I think we're gonna be a long ways from the generic, the internet version or public consumption of information that's, that's fed into G Gen ai. 'cause we, you and I could manipulate that all day long by producing more and more articles, um, versus more of the enterprise approach where it's very consistent data, structured data and governed data.

So there's a world of difference between something that's general purpose and something that is narrowly focused on a particular task and has been trained with data vetted for that task. Absolutely. Absolutely.

So last question. What's your best advice to folks? I mean, what are you seeing that folks who are getting started doing well?

Um, is it just, you know, throw everybody in the deep end of the pool or is there some like structured way of thinking about this? Uh, I use a term that is, uh, there is going to be a tremendous amount of money in the enterprise spent on gen ai and there is not gonna be a tremendous amount of money saved in the near term on gen ai, right? We, if you listen to any CEO out there of any big enterprise, you know, they'll, they'll talk about AI instantly.

'cause that's what the market's driving, that's what the, the hype cycle cycles drive in. So in that concept, I would give advice to organizations to think about, um, the value creation. So what is the outcome?

If we get this, what really will it save? Or what is the investment to get there to save? Don't just chase the checkbox of I've got an AI solution.

Don't just chase that, that component and then spend your time right now. If you don't know what those use cases are and you haven't validated the business case for it, spend the time and the structuring and the governing of the data 'cause that's the most critical part that's not being talked about very enough, very much right now. Organize, tag, govern that data, clean it up so that when you have the use case and when you have that ability, you can take advantage of it very quickly.

All right, folks, you heard it here. One, figure out the use cases. Two, remember even in the age of ai, garbage in is still garbage out.

Hey Joe, thanks for being on the show. Yeah, thank you very much. Enjoyed it.

All right, and thank you all for watching the latest episode of the text on that AI videos here. You can find this and others on our website. We invite you to check them all out.

Until then, we'll see you next time.