Billion-Dollar Software Still Fails If Nobody Uses It

[00:00:00] Jetzt nur du, dein Podcast und eine leckere Auszeit. Ab zu Aldi Nord. Für 1,99 Euro machst du es dir instant lecker. Und für 59 Cent findest du dein neues Lieblingsdessert. Bei Aldi Nord findest du immer das passende. Klingt gut? Dann probier das Asia Green Garden Udon Nudel Bowl Fertiggericht. Ab 249 Gramm für nur 1,99 Euro. Oder Mezani Sahnekefir. 250 Gramm für nur 59 Cent. Das ist Gutes für alle zum Aldi Preis. Jetzt in deiner Filiale. Aldi. Gutes für alle.

[00:00:35] Hey, this is William Tenkup. You're listening, watching the Use Case Podcast, where we talk to founders, CEOs of software firms, and we get them to kind of talk about their software. They kind of break it down. And we have Maor on today. Would you do us a favor and introduce yourself?

[00:00:52] Yeah, of course. Thank you, William, for having me. I'm super excited today. Maor, the founder of AI.Work, founded it about a year and a half ago after a long run at WalkMe, the digital adoption platform, helped build that company, and now transforming basically how agentic AI is integrated into the enterprise.

[00:01:16] And so the WalkMe story is really interesting to me because I studied user adoption of HR or work tech software starting in 2010. Wow. So I did a five-year bid basically consulting the vendor community on, okay, the sales bell rings. Okay, what? Now, how do you get them to use the software? Yeah. So usage, consumption, adoption, how do you get them to use the software? Yeah.

[00:01:46] Because in the old days, on-prem software, you didn't really care if they used the software. They bought the software. So you didn't really care. I mean, they did, but didn't. With SaaS, it's too easy to rip out. And so you got to get people hooked. You got to get them in. You got to get them using and getting the value of the software. And when I first saw WalkMe, I'm like, it's in-app. I love it. It's my favorite solution. I think it was a great acquisition, quite frankly.

[00:02:13] And I think they've kept it relatively independent. I'm not sure, but I think I've heard on the street that, you know, Workday, Oracle, anybody else, can you still use it? Which is great. Because in application, for those that are listening, if you're logging into SuccessFactors and you haven't logged in in a year, you know, you don't even know where to start, WalkMe then helps you get to the places and do the things that you need to do.

[00:02:42] So I thought it was a genius, actually a genius play. So congratulations. In general, and like since 2010, I think what happened is everybody sold all these like on-prem and even the SaaS solutions with just a technical buyer in mind. Okay, this is what he needs. This is the functionality. But then nobody thought, hold on.

[00:03:05] There are a million employees who actually need to use this application and the whole thing is going to fail if we don't get them to use. So what does it matter if I created 50 million features for some business leader and nobody uses the app? And that's where WalkMe came in and really built a strong digital adoption platform. Oh my God. And the hit with SAP is just a perfect fit. It is a perfect fit.

[00:03:32] So you did that for a while. Then after that, kind of wrote that out. AI.team, agentic AI. AI, sorry? AI.work. You know, I knew that because I was thinking about something else. So AI.work, agentic AI, and that's workflow related, right? So you can then create in town acquisition or in compensation or wherever you need to be, right?

[00:04:00] So tell us a little bit of how customers are using this offer right now. So maybe I'll even give a little background. Today, 45% of internal workflows are service oriented. So if you think about it, tens of thousands of workflows are, hey, X department, I need something. And what's happening is it's human to human. Hey, IT, I need access to this piece of software.

[00:04:27] Hey, legal, can you review my customer red line? Hey, procurement, I need to onboard a new vendor. Hey, HR and so on and so on. Oh my God, yeah. And all these, hey, this is what basically makes a company productive or not. And as you scale more employees, more employees, they just keep throwing more humans into the mix and more software trying to solve and facilitate a lot of these transactional workflows.

[00:04:57] And I think what's changed is with AI, we're finally at a point where, this is interesting, it's not this clear divide between software and humans. It was always humans operating software. And now suddenly AI can think, hold on, can it operate the software? Do I even need the software? Can it operate the workflow, the business logic? And that's where we come in.

[00:05:27] We're super focused on internal service and trying to basically break down the barrier between human and software and really leverage native agentic AI. I know people say it, but in reality, they don't really do it. What we did is we built a full-on, non-deterministic agentic platform that can handle complex processes like a human would.

[00:05:56] Basically, we look at what the humans do with the software and we mimic that experience and we build AI workers. And so, are these off the shelf based on a problem or based on a service that you've kind of already seen happen and then you tweak them to the organization or are they custom-ish? So, I would love to tell you everything is off the shelf, but in enterprise, nothing is off the shelf. No, of course.

[00:06:25] It's got to be tethered to how they do things. So, here's how it works. The AI worker connects to all your tech stack. We have a full integration platform, Magentic, connects to your tech stack. So, let's say you're working on ServiceNow and using SuccessFactors and you've got all the great enterprise applications. Now, in reality, a lot of these service workflows are operating those apps, right? So, the AI worker connects into it. It's triggered by a message that comes in.

[00:06:54] Hey, William is asking for access to Salesforce. Okay, now as a human, I would go, who's William? What does he do? What's the policy in the company? I'd read the policy. Hold on, he's a VP, so he's entitled. No, he's in marketing. He can get that type of user. And then you'd work down that workflow. Do we have licenses? What does it mean? Do I need approvals for managers? There is a whole world of process in between.

[00:07:24] And we try to give you the first 70%. So, we try to build out the templates and learn from all the customer builds and kind of replicate it with other customers. But in reality, everybody has a different tech stack and a little bit of a different policy. So, I would answer it's 30% to 40% custom for the usual use cases.

[00:07:53] And we're seeing full-on custom use cases. Right, right, right. I've said in the past on stage that I don't think any two companies in the world have the exact same tech stack. Yeah. And even if they do, they added 50 different custom fields. That's right. That's right. Messed around with it for 15 years. And now we get it, right? Yeah, yeah. Yeah, yeah. So, Argentic, a couple years ago, was more thought of, at least here in the States, as wayfinding.

[00:08:23] So, I need to know the vision plan in Indiana for, you know, as an employee, what's the, what's the bit? And it would route you to the place. And then you can find a PDF and do your thing. But I've always been curious about, okay, does it stop there or does it start recommending things? Oh, it does much more than that. You know what I'm saying? Like, but back, again, this is probably 21. A couple years ago, it was just like, hey, we can help route you to the place.

[00:08:53] And so, get your employees, get it out of the HR business partner's hands or get it out of the HR specialist's hands and just get it into, they just need a doc? It's over there? Yeah, we can do that. Let me get it to you, right? That was what the entire world did in the last 10 years. If you look at automation or even WalkMe or UiPath and these great companies, it was very if this, then that. Right, right, right, right.

[00:09:22] Okay, we have a process. You go here, you grab this document. Oh, you're onboarding a new employee. This is the process. This is what they need to do. But I think what we've seen throughout the years, look, at the end of the day, we've been doing it for a decade. So, why do we need a genetic AI? Why is this like, why is the world going crazy about AI? And the reason is it didn't really work. It was very rigid. It broke every second that something changed.

[00:09:52] You need to hire 500 service people to maintain it and make sure that every small button that changes, the whole workflow of onboarding blows up. And it just didn't deliver the real ROI. But more so, it couldn't break away from the software. Now we're seeing it break into the human side, right? That's why.

[00:10:19] Second is, I think everything was like, take service now. They try to structure everything. Here's a form. This is how you ask for a request. Here are the required fields. Here is the drop down. This is exactly what you think. You need every single edge case in that tiny form, which is obviously not the best employee experience. Sorry, it's service now. No, no, no. It could be any, you could have said any vendor. Yeah, any vendor, right?

[00:10:49] It doesn't matter. Forms were the popular. So it was all forms and if this, then that logic. And now the world has moved to an unstructured way, right? So I want to just type in Slack. I just want to say, hey. Or, hey. Or exactly, hey. Can I just fix this machine that's not working and I'm in the point of sale and I'm in the middle of with a customer and now it's bringing up a pop-up. What do I do, right?

[00:11:17] So where AI shines when you really look at agentic AI is one, understanding the unstructured request, which is a beautiful thing. Like I could show you a million examples of like people writing the most weird tickets and asking for the most vague things. Hey, I need access to the dashboard for my next QBR. Who are you? What dashboard? What are you talking about, right? And that's where AI one breaks it apart,

[00:11:47] understands it, learns from the actions. And we can touch about learning because it's a whole thing we've gone into. And it starts operating it, but it starts navigating it without the edge cases. It thinks about the edge cases and it goes to the human approval when it needs, right? And I think that's what changed. And I don't know if organizations are 100% there yet

[00:12:15] on understanding the beautiful things that could come out of agentic AI. We're just in the beginning, beginning, beginning of this revolution, evolution, whatever you want to call it. But we're seeing amazing things. We're seeing human skills. We're seeing everything. I think some of the first, it's here in the States, again, I'll talk about here, is the vendors marketed AI

[00:12:45] that was really machine learning or NLP. It wasn't actual AI. And so, you know, you got a whole kind of group of people, the prospects, practitioners, that like they're talking over their head. They're not talking about like what's the outcome. They're talking about the how. And I think it still happens too much. The vendors will just talk about the how a lot. They'll just beat people to death with the how.

[00:13:12] And it's like you care and don't care about the how. You care about what's the outcome. What do we, at the end of the day, what do we get? And I wanted to ask you about ChatGPT when it went mainstream. Did that help your kind of the cause? Did that help you all explain things or not? I think that's what unlocked the entire market. Right. You know, we came in after,

[00:13:42] we started this company a year and a half ago. So we came in after the ChatGPT revolution. But I was at WalkMe back then. And not only did it help, it unlocked like, hold on, could there be a better way? Oh, we're not stuck in like doing it the way we've been doing it for 20 years. And that's when things started going crazy, right? In the whole market. Right. So how do we get, with your background,

[00:14:12] how do we get people to use it? So I understand the sales process. We'll talk a little bit more about that stuff. But at the end of the day, there's change. There's a delta, the old way, new way, great. How do we, how do we get them to use the agents? Yeah, it's a good question. I think there's a lot of openness, especially in leadership today, of like, we, you know, we,

[00:14:41] we started out by building autonomous agents. Oh, which, and, and we realized that, we still do. And we realized that the market might not be ready for autonomous agents. Hold on. Do I trust this technology? You know, then you get into the line of business. Is it replacing human beings? Is our, you know, hold on. Are you killing jobs? Are you not killing jobs? There's, there's like, it's real, right? Yeah. I feel that every day. And,

[00:15:10] and they're right. Some jobs are being transitioned. I don't think replaced, but transitioned. And it required suddenly line of business to be a little bit more technical, more, more, you know, kind of transformational, not just knowledge workers that are doing a certain policy. And, and I think once people see the magic, that's when they want to try it to your question. Hold on. It can make me better.

[00:15:40] ChatGPT did that very well, right? I, oh, it can make me better. And then I think in R and D you saw cursor in the likes, they came in with the co-pilots and assistants and hold on. I'm, I'm coding to 10 X faster. And now we're seeing that break into customer support and now into the internal, internal workflows. And, and I think there's a lot, the market has really changed in the last six months.

[00:16:09] And I think there's a lot of skepticism, of course. Right. Could it work? Should we pull it back a little bit? And so on and so on. Um, and there's a little bit of like, is, am I choosing the right solution? My job is at risk and all that. But I think the ROI is there now. Like I can show today what I couldn't show a year ago, which is look at my AI worker in an enterprise public company,

[00:16:38] resolving already 30% of the IT tickets and look at him move in 30 days from number 10 best, um, employee that handles the workload to number one. And you see that graph in front of your eyes and you're like, okay, there's, there's something happening and I, and I want to try it. AI doesn't take breaks. It doesn't have day off. It doesn't can't unionize.

[00:17:08] Like it's 24 seven. And productivity wise, which is kind of getting to a sales question. Um, when you, when you talk to somebody, do you lean hard into productivity and just say, Hey, this is a game of just making your people better and ultimately squeezing all that inefficiency out, uh, and making everyone more productive. Like what's the bit when you, when you talk to a prospect? I think, um,

[00:17:36] I leaned in on that in the beginning, but I think, I think people are, especially leaders, their CIOs, VPs, they're, it's clear. I don't even need to make the case anymore. They understand that some of the work can be offloaded to AI today. And if it can be offloaded to AI or even assisted by AI, okay, let's not go all the way to offload it, but assisted by AI, then hold on. Um,

[00:18:06] I now am not fighting my internal battle to get more headcount next quarter or next year, just to have more people to do my, to do bigger things in the organization. I'm now fighting off to clear up my existing people and I can take on a lot more and I can bring an impact to the organization, whether it's direct productivity in the terms of like doing or, or,

[00:18:33] or the company revenue in the bottom line of how it can operate. That's what I'm seeing. Has anyone asked you, this probably is a dumb question, but I'll, I'll do it anyhow. Has anyone asked you that the investment's the same? Meaning I could do this with headcount or I could do this with AI, meaning that at the end of the day, it's $300,000. It's either I, I purchase, you know, however many employees or I purchase this in AI. Yeah. Has that,

[00:19:03] has that come up at all? It comes, the, the whole labor budget versus, software is, is a new conversation. It's happening a lot. Okay. It's happening a lot. And, and I think people compare it right now, very tactically, not strategically. So, Hey, okay, I'll reduce one employee. That's a hundred K and I'll spend a hundred K with you. And, and fine. But,

[00:19:29] but the reality is AI is even more exponential because it, it, the value, what we're seeing is the Valley grows over time. Right. It's not like you're implementing in 10 days and that's it. And that's the value you're going to get forever. You're implementing and it goes over time. So I think you're, you're even more than that. You're future proofing. You're building the machine right now. And, and yeah, we get a lot into those conversations. It's, it's interesting. I don't love them.

[00:19:58] I don't love being in a position where this is replacing people. I would actually, I actually, we added a, sorry for a second. We added a, a, a killer feature. Now, our learning module, where basically, we're basically, we have a co-pilot. So that practitioner, instead of just putting automation above him,

[00:20:22] the practitioner uses our AI worker and tells him instead of jumping apps and everything and tells him exactly what to do. And then with that, the, it starts learning the workflows. And I think, and I think people, what we're seeing is to my point, people are starting to be more AI transformation people, even like he, right? Like help desk, instead of just solving the problem, they're now teaching AI.

[00:20:50] They've learned the new tool there. They got new skills. Not only did the AI get new skills, they got new skills. You know, it's interesting because the way I fought that or way I answer that with folks that ask me is like, listen, here, here's the deal. Comes down to learning. Humans have a rate in which they can learn and they cap. So there's a speed in which they can learn. And those others, they tap, everyone taps out. Everyone,

[00:21:20] there's a ceiling. AI doesn't have either of those problems. And so it's going to learn from itself. It's going to learn the more you feed it, the more you correct it, et cetera. It, it doesn't have a ceiling and it doesn't have any problem with rapidity. It, it can learn fast. It can learn faster than any human. And oh, by the way, there's no cap. So it just, I love the idea of augmentation where the humans,

[00:21:49] as you said, the humans basically still need to be there to make sure that it's doing what it's supposed to be doing, but it can learn much faster than a human being. Like it's not even, it's not even a discussion. It also solves a more interesting problem, right? So there's, there's, we spoke about not just putting automation and replacing people, but letting the people do their job more efficiently. And then you're clearing them up and it's learning.

[00:22:18] But I also saw a lot, why we did this feature actually, is we saw a lot that organizations are wasting a lot of time doing discovery. Now they bring in a consultant or the managers and they try to figure out the use cases, but guess who knows the use cases very well? The practitioners that are doing their job. That's right. Right. So why not utilize them? This is a double win. One, you're making them more part of the process,

[00:22:48] right? They're learning to use AI tools. They're learning to build out workflows and agents and all that. And second, we're learning from the real people who are doing it. Right. And, and not wasting time thinking just about the discovery top down. So we see, we do both obviously top down, but the bottom up is really cool. We're putting, giving them the power. Imagine you have a co-pilot that sees every ticket that comes in, in service now or wherever, right?

[00:23:17] Success factors comes in a task or whatever the trigger is. And then it tells you, Hey, William, how do you want to operate this? I'm connected to all your tech stack. I'm AI. I understand everything you want. Just tell me what to do. Oh, check who William is on Okta. Oh, and can you check user groups that exist for whatever he's asking for in Entra? And can you check licenses? And can you go to, can you, by the way, ping him on Slack and tell them that we need more information?

[00:23:44] And suddenly you're talking to this agent that's performing all this, all what you need on your tech stack without having to move from it. And all the agents are in place. And it's also learning. And at the end of that process, it's like, Hey, do you want me to save this as a skill? So next time, all you have to do is just tell me to run this skill and I can do this whole thing that we just done. One, and, and now you're giving a lot of value to the company because one,

[00:24:13] you're working faster. You've just built the AI, the beginning of the AI transformation. And, and three, they don't have to do the discovery top down. You're doing the transformation. Right. Okay. So I thought about you the other day in the, uh, I read an article on business, business insider. It was about culture actually. And what it was basically, it kind of got down to within every organization is two tribes,

[00:24:39] a tribe of looking at AI and saying they're excited and they're like getting up every day. And they're, you know, they're like, they want to figure it all out. And then there's another tribe that puts their head in the sand. They just want to go away. Yeah. Okay. So the question that, that comes, that comes up for me is whose responsibility is it to train? Cause that learning part, that component about AI,

[00:25:07] is it the employee or is it the company that, that gets them to the place of learning? I think it's both. Look, there's, there's no way around it. If organizations, if we, as humanity made a choice that we're going, all in on an AI and, you know, look at Davos would just happen. Like it was all AI, right? Every leader in the world. So I think we've already made that decision. Yes. The, the CEOs and the venture money. And,

[00:25:37] and like, that's already been, that's done. The transformation is underway. Now you either get with the program or you don't get with the program. Right. And, and, and I think, the smart ones are the ones that are really, it's, it's all a game of skills. Yes. Skills for the AI to do certain tasks, but also new skills for the human beings. I learned a lot of new skills. I'm building agents.

[00:26:06] I'm doing half my marketing with agents today. It's a new skill. I was CMO of walk me. I didn't use agents at all. It didn't exist. Right. I might've used sequences and automation. Not this, not this. Right. So I'm learning new skills, even as a marketer or a founder. And I think it's up to everyone in the organization to really push that transformation. And when you look at culture, the, the,

[00:26:32] the companies that are succeeding are the ones that the management is aligned with the employees. So, so are you following this trend? I think it's mostly out of Silicon Valley right now of revenue to headcount. So they're falling in love with this, this number, the ratio being much higher per employee. Wall street's kind of starting to fall in love with that number as well. Yeah.

[00:27:02] That kind of gets us out of the business of employee headcount reduction. Cause it actually, if they, if the money falls in love with that ratio, then that ratio is going to be what people were judged on. I think from someone who worked walk me with a public company for two years before he got acquired by SAP. And I had the privilege. Yeah. We did a NASDAQ IPO back in. Cool. 21. But as someone who worked very closely on,

[00:27:31] on the stock and earnings and strategy of the company, there are always trends. There are always different metrics that we try to invent to, to, to kind of, cause otherwise it's a mess. How do you evaluate your VC? How do you evaluate if he's good or not? So they put on these metrics. So, you know, one time it's growth, then it's a profitability. Now it's a, this ratio. Yeah. So I, yeah, exactly. Tam before.

[00:27:58] So I don't think it's a cool metric. No, I don't either. And, and, and, but I don't think it's a, you know, I don't think that matters. I think what matters is speed. In, in 10 years or in five years, you're this huge enterprise company, right? Like whatever you're manufacturing. Okay. You don't have to be in the tech world, right? In 10 years, does your competitor, if they set it up, it doesn't matter.

[00:28:27] Let's say they have the same amount of headcount, but they set up a ton of AI. Are they going faster? Are they manufacturing faster? Are they distributing faster? Are they marketing faster? Then you're screwed. You know, this as a recovering marketer, it's a, it's the, it's a magic triangle of price, quality, and speed. So faster being one of those components, but also a component being quality. So you can be fast and you can be faster,

[00:28:55] but if the quality is less or the price is more, so what? So you've got to, you've got that magic quadrant that marketers have known forever. It's like, okay, I've got to, I've got to figure out this, you know, price, quality, speed. I wouldn't be worried today as a company or, or even myself, right? As our company, I'm not worried today about the ratio of AI to employees. I'm, I'm,

[00:29:20] I'm worried about finding the right product that solves the business outcome. Right. That's what matters. That's what we should all be worried about. I mean, exactly. Yeah. The outcomes, the outcomes because playing games or, or, you know, saying that you have 50 AI agents when all it is, is one, you know, one simple process. And it's, it's that doesn't impress me. What impresses me is we just, you know,

[00:29:48] we just cut down 30% of the work and the team is doing other stuff. That, that impresses me, or you don't have to wait three days for someone to reply. You get a reply in two seconds. I mean, that's an outcome. That's me working faster. Yeah. I love, I love that. Well, I love that. I love that for a number of reasons, because it's, again, I ask a question, I get an answer,

[00:30:17] but it's a great answer. Like it's something I didn't even think about. Like, Oh my God, I hadn't even like, like I was just asking a basic question. They went deeper than I was thinking. I love that. So that's the rag revolution. Right. That's the knowledge revolution. We're in kind of the next phase that the action revolution. Right. Can it also do. Right. And, and, and again, that's getting back to your speed part. Cause it's gotta be able, it's gotta be,

[00:30:45] we've gotta be able to judge it on not just, you know, how it's built and all that other stuff like, but the action layer. So what does it ultimately do? It solves this problem. That's great. And answers a question. Great. But what is it? What's the action layer? Recommendation layer is cool. However, we want to get past recommendations and then into action. Two questions before we sign off. One is your favorite part of the demo.

[00:31:15] And it's, it's kind of like when you get to show somebody something and they're, maybe they're, maybe they're a naysayer or a critic or, you know, just someone who's kind of negative, but you got them on a call and you're showing them some stuff. What's the part where you love to, you get them to this place and then go, if they can't see it here, they're not going to see it. Yeah. I have a couple, we map them out as magic points, right? Ah, no shit. All right. And,

[00:31:44] and there are a couple of magic points where you see it and just something in your mind says, Oh, holy. Wait a minute. Wait a minute. Yeah. Oh, I didn't know AI could do that. And we have that by the way, every single week, even internally, we see like when customers were working or we're playing around with the system. We have those moments where we're like, Oh, hold on. Under 10 of consequence.

[00:32:11] I didn't tell it to calculate how many tomatoes, but it actually saw that there were tomato and did that. And like these little why moments, right? Yeah. So, so when I demo and they see a ticket comes in and it just starts operating like a human would operate. And they start seeing it, it flow. Right. I click the button. It starts going, boom, boom, boom, going to Okta, doing this, starts running. And you see,

[00:32:41] we added a cool feature where you can see the thought process, similar like chat. And then you're like, and then you see the agent talking to itself, right? I need to go grab the policy because I don't know what the Monday users are. And then he goes in and now I need to check the app list for Monday licenses. And he goes somewhere else. And now I've come to the conclusion that it needs this and this. When people see that piece in the demo, they're like, Oh, Hey,

[00:33:09] this is not automation that I've seen in my life. Right. This is, this is a, I got chills. That was great. Yeah. Yeah. I got to show you that demo, man. You'll see. Oh, well, it's a customer build, but yeah. Well, okay. So the last thing is where you are having this conversation a year from now, not too far out, not flying cars, but just a year from now. What do you think we're talking about? When are we getting to the flying cars? Dude, I'm down. I'm down. I'm ready.

[00:33:38] If you get one, tell me, cause I'm down. I'm not afraid. Let's go. Where are we one year down? I think, I think just from our perspective, I think people get autonomous AI. I think that's a learning curve. And it's a trust. That's a trust thing, right? Trust thing. It's an experiment thing. It's a maturity thing with the AI too. Right. Yeah. And the AI gets much better. And,

[00:34:07] and I think I'm getting phone calls of like, I want to hire an HR ops. Worker. And they're not thinking about, I need AI to do this. I need to do transform. I, I need work day to go from here to there. Right. I just need this solution. Cause I want to drive this outcome. And I think that's what we'll, we'll start seeing.

[00:34:32] We'll seeing mature solutions and mature market asking for those solutions. Drops mic, walks off stage, brother. This has been fantastic. Thanks so much for your time. How can, how can folks find you? They can go to AI.org. They can ping me on LinkedIn, or they can just email me, maor, M A O R at AI.org. Always happy to connect. Always happy to speak with everyone. Done deal. Thank you so much for carving out time for us, brother. Yeah. Thank you so much.

[00:35:02] Appreciate it. Thank you.