KJ Kusch on How AI Agents Are Transforming the Application User Experience

Hello and welcome to the latest edition of the Techron AI Leadership Insight series. I'm your host, Mike Bazar. Today we're with KJ Kusch, who's global field CTO for Wfme.

That's in unit of SAP, that focuses on optimizing the end user experience with a framework they developed. And we're gonna be talking about that in the context of artificial intelligence 'cause it's the age of AI agents. Kj, welcome to show.

Thank you Michael. So I'm not sure everybody watching this knows exactly what WalkMe is. So maybe let's start at the beginning and describe what is this end user framework and how come we need it?

Because well, we've been accessing applications for better or worse for years. Yeah. Which is true.

Right? And if you think about it, everybody shows up with a large ecosystem with lots of applications. And what we've seen over the years is best intent on billing applications, users have to use it.

And that point of friction is where adoption comes in. So we did training and then we went to the application and used it WalkMe, combine those. So what we're really focused on is actually helping users navigate, adopt, and execute process.

What's interesting in the age of AI is how much AI is working its way into process. So our worlds are now combining AI with basically adoption of that technology. We are also on the cusp of the age of AI agents and SAP has Juul, but there's a bunch of other ones that are out there.

How will AI agents kind of get streamlined into that user experience process? Because I don't think AI agents will be doing everything and somebody's gotta orchestrate these things. The one thing I could tell you about working with so many customers is they have a hybrid environment.

They use multiple technologies. And really AI needs to be orchestrated, not just enabled, just not just turned on per application. And that's what we're finding with the users because we focus on that.

So if somebody is guiding, and I'll, and I'll give you an example. Assistant to me has a system process. SAP wants you to use S four hana SuccessFactors in a certain way.

Then my customers, everybody out there in the universe says, here's really my standard operating procedure. Here's how I go across eight applications to do that. And now I have a Gemini and I have a, you know, Juul all trying to tell users what to do.

But it's disconnecting more and more that what the user is trying to do by giving spurts of AI information. And so now when we look at where WalkMe is really sitting in a day in the life, it's more of orchestrating that ai. So we are right, unlike SAP focused on the SAP environment, right?

Or Watson on the IBM or Gemini in in its own is we look at it across all of those and not from the viewpoint of that application or right, or that that ecosystem, it's really agnostic and that's where we stand out. It's not being a competitor against those, it's being a force multiplier for those so that users can start orchestrating across them contextually based on where the user's at. 'cause ultimately companies just need to get work done.

That's how they make money. So to your point, I might be using two different applications that have two different sets of agents and I'm asking them to perform a task, how will they negotiate with each other to divvy up that task and have that conversation amongst themselves and what happens when they get in an argument? Yeah.

So I don't think it's necessarily going to be like two suggestions provided at one time. And here's why I say that because truly, right, the underlying application knows its business object and its data. And it's going to say at this point, I'll just give you an example.

You're doing a purchase order, right? And based on what I know from what you put in the value of say this order and these line items, it's gonna suggest discounting. So nobody's going to argue with the discounting that comes from that core app.

What's gonna truly happen for that user says, sounds great, except my internal company policy, right? Might conflict with that. And so in their brain, not on the screen, there's a point of friction.

And so what we see at WalkMe is we're watching and there's a long pause and the next thing you know, there's a workaround. They go look at their knowledge article and they come back. So now what you have to think about is how can WalkMe help?

Well, right, it, it is a matter of, now I need to integrate right this, this what I would call a re, but it's really a point over to that customer's, right? What I would call standard operating procedure to say, yes, this is correct or not. If I have to do say shipping right inside of S four hana, how do I know it's written and regulatory of this country, this location, right?

That call off to another source of information is literally should be sequenced correctly for the customer, which is why WalkMe is the right orchestrator for that. Because what's happening is there is the system process combined with a customer's, right? What they call their user flow or their workflow.

And WalkMe combines those to sequence them and bring in the source of the decision point that's needed. So we're not gonna, right, if a user's not frustrated and they can follow the process and, and a JUUL or a Watson or anybody can do it, we're not gonna insert. But if we see frustration or we see slowdowns or cycle time, right?

We wanna start proactively prompting or prompting, right? Hey, I have an idea or I see a frustration or I've seen on the screen you've gotten it wrong three times and it pops up the right source to answer that point of friction versus trying to reorchestrate the whole process is orchestrating the process. We're literally facilitating the user workflow.

Will there be more of these bottlenecks in the age of ai? Because we will, I'll have say my set of AI agents and I may have optimized that to, for example, sell something and we'll be engaging with customers who probably will have AI agents that are optimized to procure things. And these two things will have to kind of negotiate with each other.

But one, I wanna surface something when they're just basically kind of grounding each other to a halt because they're so well optimized against each other that some human intervention may be required. Yeah, and it, it's a great question. So I think, right, AI had this really high swing where everybody felt like it was very productive.

And then I, I know the whole entire AI community is now saying, yeah, but let's step back and ask, are we hallucinating? How valid is the source of data? Right?

And, and so what's interesting to me, the trend over the last probably three years is while at first we thought it was great, now customers are questioning it. So I think what's gonna happen is companies are starting to become aware of the data results that come from ai. And so that is a point of friction.

It's actually you're watching users slow down and question. Now some AI sources the data, some doesn't, right? And even sourcing the data is interesting because is it my company's right answer or is it the general from an LLM or PLM or an industry LLM, right?

And so what's truly happening is, I think friction points. Like I think what's happening is people are gonna want that really fast cycle time that comes from ai and then they're gonna start looking at the visibility, the quality results of what they're doing is doing it, you know, 90 times faster, better than doing it 85 times with better quality. So is right, is the audit results of what's the audit results of machine learning AI versus humans And how many times do I interject?

I suspect we're gonna see another bump, which is in this cycle of ai, which says, okay, I need to insert, I am seeing, you know, incorrect errors, I'm going to insert some ai, right? I'm going to see AI insert and the results come back wrong and we're gonna see interjections of the right AI for the right moment. And then maybe in three to four years it's gonna be smooth again.

So I think we're gonna have a couple bumps when we think about, you know, what's the right data, how do I source it? Is it frustrating the user? I think we're gonna have some, let's try this, let's not, let's check the quality of data.

And over time companies will figure it out. That's how they're gonna get to the ROI for, for truly for ai, it can't just be fast, it has to be right. You know, I'm old enough to remember good computer science and we used to separate the presentation layer from the outline code.

And um, and somehow I feel like we got away from that over the years and we wound up customizing things and tying things too closely together. So are we now gonna revisit all that because we're starting to understand that we need to separate the workflows and the presentations from the underlying code. I think you're correct, right?

Because we, I'm, I'm old enough to say, right? We had a couple apps and then we bought a whole lot of apps and then we said, oh, we gotta do apps consolidation, right? If I just have to say it's like for, like, it pretty much is like for like where, right?

It may not be just the number of apps, but the user experience and as well tended as the best technologies can be. If users have to use more than one application, and we know that's true most of the time, then we have to really start talking about that user experience layer. Can it really be controlled by individual, you know, companies like SAP or do I need a WalkMe to really drive that?

And of course I'm at WalkMe, so I believe we need that common experience because you know, as well, I, I mean I've been a developer in my lifetime and I thought I did it right, like it, it, I think it was easier to say, I developed my technology, I'm good to go. And I never had to consider humans. I think humans are the reason we have to start looking at that whole presentation layer, right?

And does it truly need to be one app at a time or can I make it universal? And with AI now we have this whole new debate, is it really a, you know, let's commonize the screen so that people can get through different screens Or is AI now the common presentation layer? Can I have fact turn everything in the backend to just the database and everything can be ai, maybe it's just a matter of when, right?

And what that experience is like and how are users really gonna become prompters instead of followers of applications? I mean we've built applications for so long, that's what we're doing. So I think there's a lot of human factors, education factors, training factors, right?

Is this for the young generation, not the new generation? Can I afford to do both in parallel? Um, I think people are gonna play around with all options.

Is there an opportunity to maybe consolidate a lot of these applications that we've had? And that's become a more pressing issue in my mind in the post COVID era. 'cause during COVID, we seemed to go out and buy a SAS app for every little workflow function we needed.

And we were just like, you know, maybe be a little bit like drunken sailors, but we wanted to get things done and devil be damn what it costs. And now I walk around and I'll talk to people and they have hundreds of SaaS apps with overlapping workflows and stitching that all together is become somewhat problematic. So, or we on the cusp of some other wave of consolidation and rethinking workflows.

I think we are. So let's just, just talk about right, how data has changed that ecosystem. So for the most part, like in my history, I've bought a lot of software and the reason that I bought software is 'cause it served a niche, right?

It it was my industry focus, it really drove my workflow. And instead of having one application where it did one thing well and, and nine things kind of well, right? We, we bought the big one, the, the Gartner leader and then I bought nine ancillaries.

But now with ai, the way data and AI and machine learning is happening, I don't know that I need the application as much as I need the data and the smarts that I got from it. And then I can, I can bring it back to less applications. That's how I see tying it together.

If the data, the objects and the machine learning can give me the niche process I need to make my company unique, then I don't need the whole application, right? I just need what AI can produce from those right. Custom experiences and the reason I bought.

So absolutely. I also think right with WalkMe, we're very fortunate that we have a tool called Discovery. And what we're seeing is are people really using all of these applications we bought?

And what I mean using, I don't mean, right? There's a lot of discovery tools that can log in, say somebody used it and log back out. What I really care about is how they're using it.

Are they logging in? How long are they staying in? Right?

Are they doing five minutes of work and then logging out, right? Are they taking that work from that niche application back to a bigger application? Is this supposed to supplement or is it supposed to be standalone?

Right? And I think how people use all of those applications is driving the need to just, I don't need them anymore. Right?

You said you needed this, I'll just say a fancy project tool and in fact you're not doing anything different than the license we have over here, right? You're you what you said you wanted it for, you're not actually using it for, I can tell, I can look at the screens and tell you you're not, it's going away, right? I think a lot of business cases never got traced and we don't use the reasons I bought 'em.

I think WalkMe's gonna help really downsize that ecosystem by saying here's what people really are doing, right? It's an answer we never had in the past. What's that one thing that you see organizations doing that just still makes you shake your head a little bit and goes, you know, folks, maybe we could be a little bit smarter.

So I think a lot of people are still heavily reliable on manual work. I'm really, really amazed, right? That, um, they haven't even used AI to automate manual tasks.

I, I look at like, and not in it, I think it users and back office can, but when I really look at what's business critical, like I'm, I'm manufacturing, I have to get stuff out, right? Um, I'm surprised how much people are still documenting, they're still using, like even on their phone, right? There's no AI on simple transactions.

They're not even starting small on some of the business critical parts of their company. And I'm like, that part of AI is safe. I believe that data, that machine learning, like I can predict what a delivery code's gonna be.

I can tell you how you're gonna have to pick and pack. They're still manually entering everything, right? So I think I'm, I'm more amazed that people aren't starting small and then all of a sudden customers are getting this pressure and I, I look at customers, they're like, okay, now I need to solve world hunger with ai.

And they haven't even done a tiny project. So the leaders started with tiny projects, right? The the innovators and the thinkers.

And now I think everybody thinks that's easy, right? And they're gonna take the lessons learned from those leaders and now all the followers are trying to do Big Bang. I think followers, people who are not natural innovators, um, also need to start small in this, in this space.

Alright, well folks, you heard it here. The user experience is changing and hopefully from the better in the age of ai. But you gotta take a minute to think about it 'cause it's just not naturally gonna happen without some actual effort and a plan.

There you have it. Hey kj, thanks for being on the show. Yeah, great.

All right. Take care. And thank you all for watching the latest episode of the Techstrong AI Leadership series.

You can find this episode and others on our website. We invite you to check all those out. Until then, we'll see you next step.