#223 - EQ Is Beating IQ: Ilan Man (Paradox Machines) on Leading in the AI Era

0:00 I often tell my customers, buy something cheap off the shelf and let's just see what happens. The more you live in the world of pain, when when we talk again in six months, you'll have a very clear view of what you want to fix. Welcome back. ~ you know, you just know at the beginning of some of these podcasts that you're just gonna be out brillianced. I know all my guests, they all look like that compared to me, but today might be another version of it because today's guest is an Ilan.

0:47 It's spelled different, but Ilan Man. He's the founder and CEO of Paradox Machines. It's a data and AI company helping mid-market organizations turn fragmented data into trusted decision ready infrastructure. Y'all, my mind's already blown. He's already outsmarted me just from that description. But Ilan started his career as an actuary before moving into data science at Squarespace. We dropped these names because it starts to give us some relevance about the world. Then he later held data leadership roles at Paperless Post and Formation Bio. He's getting smarter, you all. And then he served as an executive at the Brooklyn Data Company, the Through its private equity acquisition in 2023. He holds a bachelor's degree in statistics from the University of Toronto and a master's degree in statistics from Duke.

1:30 It makes me think, what's he doing on the show? And then his argument, though, is a contrarian one in the age of AI hype, which I am so pumped about getting in today, that most companies don't have an AI problem. They have a data problem. Ilan Man, welcome to the show, man. Thanks, Mark. really appreciate that introduction. You've set the bar quite high. ~ so I gotta do some stretches here so that I can I can even touch it, but I'm up for the challenge. I think we're so we're gonna do middle stretches because I already tell you ripped and in great shape. You've got that down. So here come the middle stretches. With a guy that is sometimes you just meet these folks and you just know there's brilliance of traditional education in their mind.

2:14 But when I also meet you, I'm like, he's got a personality too. So you have a combination that is rare in your space of work and the degrees you have. So that's why I'm so interested not just to talk to you, Dave, but follow what this guy's gonna do. Because so many people that come from where you come from, you have an extremely high IQ, but the EQ in connecting socially isn't good. You don't have that problem. Well, I appreciate you saying that. I mean, let's see how this conversation goes and then you could reassess. But I actually think EQ, ~ or whatever we wanna kind of call this notion of interpersonal skills and just like getting to know people and that kind of like instinct that you have, which by the way I think can be refined and honed over time. That's just obviously becoming much more relevant and important.

2:57 I've I I read somewhere just just yesterday or the day before, as you know, all the AI doom and gloom has been happening and ~ and in particular how AI's been solving these like very crazy math problems. If you've been following, I'm not sure, but we can talk about that. That your your IQ as a kind of differentiator of your ability and then kind of a indicator of some sort of future success, which has always been kind of like basically true more more or less, is becoming less and less true in the future. And this individual didn't say that, but I think the implication is EQ, if you want to just like simplify the world into the two, that's gonna be so much more important. I think it's already more important, frankly, but I think it's gonna be even more so.

3:37 With a Did did you intentionally work on it or is this who you were? 'Cause there's this argument, do we find folks or do we help build them? I think there's you know, there's always some there's like raw material, I would say, there that was honed and and shaped and kind of the edges were were were kind of sanded down over time, for sure. As I aged, as I became more confident in myself and kind of you know, the I don't want to say I've shed the ego. I'm still very vain in many ways and and competitive and all the rest of it, but as I've shed some ego, as I've learned, frankly, as I've done more sales, you know, I I traditionally not as you kinda noted, not from a sales background, very much. Numbers, spreadsheets, code, math.

4:19 I mean, that's still my happy kind of happy place. But as I've been thrust into the sales position from my last place, and then leadership, managing people, managing and people come with emotions and baggage and all the rest of it, to be effective is to understand who you're talking to on the other side. And you know, you kind of like have to learn. Or or you just are not going to be successful. And that's fine. I I I chose to try to be successful and as a result, had to kind of stretch myself in that way. Data is our four letter word and you love it. Yeah, yeah, I love it. It sounds weird to say. I mean, what do I love? I love data. Like that sounds so odd. I'm more interested. I mean, data is an enabler of a thing.

5:00 So I'm very interested in taking again raw material, tools. Data is a, you know, is a tool and crafting it, structuring it, you know, augmenting it, and then enabling people to do what they want to do with that data. I I'm a big believer of things like the scientific method of making decisions. based in reason and fact and logical thinking and all the rest of it. And data is not the only, but it's a very effective mechanism to arrive at, let's say, a better conclusion or a better outcome. Mm. You have to be a first principles guy then too. You are Yes. So v very big first pr well, so and you gotta do both. You can't recreate the universe every time you have a question from first principles.

5:43 So you need to both like also think in analogies and ~ I would say like the thing that I'm you know that that I'm I'm most kind of I I I spike in is the most boring thing in the world, which is moderation. And I think like you need to hold multiple truths at once and and a and and and in particular know how to deploy them and like how to pull back. So First principles thinking, very important in certain domains. And in other domains, you know, somebody's done it before. Don't reinvent the wheel. Let's just keep moving. and are you like a stoic? Are you a Marcus Aurelius, Confucius, Stoic? Like are you all those things combined or something else? I I'm sure I'm n none of those things because I tried and I just always fail.

6:24 But I certainly in principle, at least on the stoic side, certainly try to like think of myself as one or adhere to a lot of those principles of separating yourself from the thing, of you know, not stoic in the like stone faced kind of like vernacular, but ~ and just, you know, like exerting control over yourself to the extent that we can, I think is really important. And even a few basic principles will go a long way, like when you you know, what falls out of this stuff is is like a way of being, which I try to do, but you know, sometimes I don't. Sometimes I lose my my stuff. So in this cra since since AI's here and it's on everybody's mind and all the leaders are getting together and they change their story every week on what's happening and what's not, and who's gonna be a trillionaire and who's gonna go public and their positioning and leveraging this and who's correct and who's right and Anthropic said this and you know, Altman and Chat said this and you know, and then Meta's doing their thing and they come out with Muse and that's what the hell's going on, Ilan?

7:30 this is this is what drives me kind of up the wall because so much is happening and what I what it what I tell folks who are not in this space is you gotta s th there's kind of like two camps or two two ways to think about it. There's AI as the tool, the chat GPT, the claw, the maybe the agent that books you a you know, a travel thing, whatever they're doing these days, sends you alerts. Then there's the the e the you know, Elon with an E, the Altman, the Dario, the kind of the trillionaires. the Business of AI. the business of AI, which is just out of it's crazy. It's out of proportion. And the two are related but separate. So and and I think it's important to to recognize that when they're talking about like AI taking over the the world and taking you know putting us out of a job and it's with trillions of dollars and all the rest of it, that has to be that like that's a narrative that they're pushing.

8:26 And you know, you don't even have to be a cynical, like you don't have to be a cynic about it. You can you could be a realist. They're out there to make money for their shareholders and they have, you know, their own perspectives and motivations. That has like a little but not that much to do with your day to day life and how you're gonna use AI. And you have to separate the two. ~ unfortunately the media, you know, they work in sound bites and they like to scare people. ~ the leaders of of these companies also like to scare scare people, and that's their MO. And I think it's important to distinguish between the two. Not get caught up in the noise. So you're you're building paradox inside infinity. That's right. Infinity's our kind of parent company.

9:07 Right. The AI services holding company. Right. So rather than taking the classic venture back startup path, what does the holding company structure let you do that a V C backed company couldn't? And but also what's it cost you doing it this way? Yeah, yeah. So I I would think about it less of what they let me do and more what they enable me to do. What they enable me to do is not worry. So they provide shared services. Legal, finance, accounting, HR. You know, there's like kind of a PR platform team that helps kind of connect me to people and send me to events. That's stuff that I I don't know the VCs that. Maybe they do, ~ but they do that as a kind of as a service as part of like joining the platform. Obviously, there's an an investment.

9:51 There's a number of other CEOs within the portfolio and I swap ideas with them and we have kind of like monthly sessions together. We have an off site. And that's been that was really important for me. I I have never started a company before and I wasn't sure what I was getting myself into, but having this kind of like safety net was really nice at this point in my life and for what what we needed. What you have to give up is half the company, which is a lot. But ~ But for what I wanted and and there's some nuance to the terms about the funding, it's non dilutive. It's it's pretty good. If you like read the fine print, ends up being like, that's a pretty good deal. And ~ but but like per personally speaking, I wasn't I wasn't about to go and v like fundraise. I don't have the means to bootstrap at the speed that I wanna bootstra that you know, that I wanna It's probably necessary to have a relevant company in this space.

10:38 It's it's it's necessary to be relevant to be able to move fast. You ni you know, if I'm like Ilan in Brooklyn, you know, like if you're I don't know, five, six hundred million dollar revenue company, you might not work with me because like as much as you like me, there's a risk factor. But having the infinity platform and then having the funding and then being able to build a team as quickly as I have been gives you this credibility, you know, and and that was that's been really helpful as well. So it seems like your model is like a serv is services plus platform, like consultants who also ship product, right? It's A hundred percent. yeah, and so the the classic critique of that services don't scale, which is why VCs a lot of times won't fund them, is what I've found, right? Make the case that the hybrid's the advantage, not the compromise.

11:27 So it's definitely the adva so so that is like the classic way of thinking. Classic like five years ago, you know. Yeah. So give me the give me the new one. Give me the new I knew it. See, I told you in the beginning this guy's gonna bring brilliance. So there's services and they're SaaS, if we wanted to like simplify the world. And you know, they come with their margins, the like twenty, thirty percent margins over here and the eighty to ninety percent plus margins over there. And you know, and and and so on. so it's like it's napkin math. You can un you can figure out why one would be investable depending on your risk tolerance versus another. In the middle is we have this new beast that has sprung up, spawned, if you will, ~ called AI services. Now, I I I'll I'll preface by saying AI services is something like a couple of years old in its current form.

12:13 Palantir might say, we've been doing AI services for a long time. But and you know, debatable. But like the new the new breed of of services companies are AI services, AI native services, of which some of ~ that's just like AI native in name, but they're like an old company and they bolted on, you know, a chat bot. And some, like mine, just started seven months ago. Like we have we have no no no option but to be AI native because like that's the tool that I use. And so these companies, they're not going to be quite, in my opinion, quite SaaS margins, which by the way are going to be decreasing anyway, but like classic SaaS margins, they're not going to be there. But as I understand it, talking to some other AI services leaders, you know, the the margins you're expecting are fifty, sixty plus percent, which for services companies at scale is really nice.

13:02 And we happen to be north of that, which is really good. Early days, not trying to you know, count my chickens too too soon. But long story short, AI has enabled you to deliver services faster, cheaper, better. Which is fine. And and and and you know, that in and of itself doesn't create a successful company. You have to then like actually execute and you have to bring in customers and and and You gotta sell. you're gonna sell and it's all about relationships. So it can't be a race to the bottom. Like if your entire business model is like cheaper and faster Well, then somebody will be like slightly cheaper and a little faster. And then, you know, we're we're kind of raced to the bottom. I certainly don't want to do that. I'm in the the the business of partnerships, of relationship building. When people buy from us, you know, I tell them out ~ out front, like, there are people who can do what we do.

13:50 We happen to be really good at it. And, you know, we have a small team of very seasoned engineers and and myself, you know, so what you're getting is not somebody who can code faster or who can use an agent better than somebody else. Or who's cheaper because we're not cheaper. It's reliability, it's excellence, it's that that trust factor. And now that's that was true before AI entered the picture. So I'm always I'm always trying to be wary. It's not that's why sometimes when somebody slaps AI on top of it and they don't see the return, they're like, what happens like, well, AI is not gonna AI by itself is not gonna do the thing that you want it to do if you know how to use it, and you have all this other stuff, which is not AI related, usually on the human element, then maybe you got a business.

14:32 So er AI services, I sat down about me being a business owner, aside from the podcast and my businesses. I forget where I was, but I was flying to Miami, I believe, to buy LED walls. Before I'm gonna spend seventy five grand or a hundred grand on huge LED walls, like I'm going to the showroom to see this stuff, right? To make sure because we do the installs. But I had a guy in town who also knew a Russian guy named Vlad. Who was in AI as a programmer, and I said, I have ideas, I want to sit down, not being the A AI guy. And I said, There is an opportunity here to help tens and hundreds of thousands of companies build proprietary agents that are gonna help them run their companies.

15:14 Because I'm speaking from an entrepreneur. And I said, You all, there's so much margin, and there's millions on the line for companies that would help do it. And I'm like, I think that's what Ilan's doing. Yeah, I mean I th so we're we're enabling mid market companies. So companies that are, you know, they're they're generating a bunch of revenue, they're probably reasonably successful, typically a little bit older. So these aren't companies one or two years old or a little bit older. From industries that I would say like non sexy industries. I mean, maybe they are, I don't know, but like traditionally non sexy industries. And we're equipping them, we're we're giving them the ability to compete at the at the highest levels. That's it. So in the past, they couldn't get a data engineer.

15:56 A VP of data, get a data strategy, buy a Snowflake warehouse, get an ETL tool, get a BI tool, connect it all, manage it, and then start finding value. You're half a million dollars in, if you're lucky, annually. And you're like, what am I what am I even doing here? Like, why like remind like we've lost the plot. So they're like, you know, scrap it. Or they tried it. Five years ago, they did the big migration from system A to system B. It was a mess. Took them 18 months, cost them $2 million. Now they're managing a new system. And they're like, I don't really, hmm. Some CIO told us to do this. And then when he left, we're, you know, stuck with this thing. And it's a very classic like enterprise example and in some bid markets. That's just like the way it worked.

16:40 Then AI came along and it said, you know, you're trying to migrate from A to B. You could just do that in a couple minutes. Yeah. yeah. A couple hours in instead of a instead of a few weeks. You're trying to. Connect these tools, I'll do you one better. Let's build the tools for you. That's it. And managing it becomes much easier. That's it. And so on and so forth. So a a lot of organizations are kind of coming online to this notion of we want to get involved with AI. We're not exactly sure how, but we're we're sold. AI is good. You know, A AI could be useful for us, you know, beyond ChatGPT and Cloud Enterprise. Like we can we can actually like integrate it somehow. What that exactly means is a little amorphous for a lot of folks.

17:22 So they look to people like me to say, hey, Ilan, we we think you know what you're talking about. We basically trust you because maybe I was referred to somebody. Take us on this journey with you, you know? And and that's kind of going back to the earlier point. Like that's that's the game I'm playing, is the, you know, let let's help get your company online on board to AI, to better data, to better decision making. So you could finally have this thing that was always elusive, always just a little bit outside of your grasp. Told them in Miami, I you all here's how I envision this. And I love it if it's what you're doing and it's the space you're in. Maybe I can speak to the layman average business term and not the brilliant data scientist that we're speaking here. They let him do his work. But I said, I see this as letting the mom and pop in the family business that has good revenue, but they just never really had the catalyst that was going to let them compete with the company ten times their size.

18:15 Not the marketing department, not the infrastructure. And if you partner with an Ilan or other people who are going to build this out, they will be able to build agents. That's the benefit of AI, how I see it for these companies. Let them build agents that are running these things when you previously couldn't have afforded, didn't have the time, because you're working on the business. They now can compete with the folks that have the outsized revenue. And what you'll find is their businesses actually might be better for the customer to interact with. when they can finally scale and use these systems at a fraction of what it used to would have cost them doing it a traditional way. That's exactly right. That's exactly right. So AI has brought the cost curve down and the feature s like the the the amount of features and what you can get out of it up.

19:01 Right. So in the past you had a box, I call it the SAS box, and you open up the SAS tool and like you drag and you fill out the form and it goes over here and that's just what it was. And if you didn't like it, you can hire a developer to augment it and that's a whole mess. So you just Or if you didn't have the budget. If you didn't have the budget, whatever it was. You know, or you download to an Excel spreadsheet and then you're manipulating and you're like, Why am I is this how it works? And then it turns out that's how ninety-nine percent of companies do it. a hundred percent. Who are we kidding? Everybody does it this way. Yes. But and and and I and I'm real with my customers, I say, like, I'm not here to take out Excel. Excel's gonna outlive us all. But your Excel, my guess, is doing a lot of work. You have a lot of tabs, a lot of formulas.

19:43 Takes you five minutes to up update the thing. And it might be wrong and you're kind of like afraid to touch it. And if the poor analyst who put it together leaves, you're like, I am really screwed. My business runs on this Excel. Okay, Excel's great. It's doing too much. You've you've you're you know, it's f it's too flexible, you know, if that makes sense. So that you can cram a lot of logic and data and processes into it. So what you know, what I propose is hey, let's take out of Excel or your spreadsheet all the things that don't need to live there that you've kind of overengineered. And let's just use it for the things that it's good for. You know, your income statement, you know, your maybe a scenario plan.

20:25 Maybe you want to do some back of the envelope calculations. Like it's really good for some stuff. All of the it's a tip of the iceberg analogy. All the things below the water, that should be done in software, in the cloud, in aware, you know, there's all kinds of systems up there. And the tip of the iceberg is your Excel sheet and it has like clean data, it's already standardized. And it's kind of like a pleasant experience actually, and not a n mind numbing manual, you know, thing that you do Friday mornings. What's gonna be the thing that makes your business succeed? Because and I think about that as your no lock in technique. Okay. So you do a no proprietary lock in.

21:05 You own it, it's a great line. It's fantastic, right? It is. Even if it's true or not, it's a great line. It's true. It's true. I I I believe I'm assuming so. I wouldn't have brought it up. Yeah. Yep. Yep. But lock in is how software companies used to make all the money. That was that was the SaaS, that was like legacy phone, it was everything. Mm-hmm. Are you leaving money on the table on principle? Or is the anti lock anti lock instance that actually the better business? Like how are you going about it? Are you building the id idealistic better business? Or do you actually think this is the better way to build the profit business now? I'm hoping that those two things are self-reinforcing.

21:47 So I do think it's a better, like in principle, it's better business. I want people to own their data, own their their systems, their logic, their AI. It's this data sovereignty, AI sovereignty that's was popular a few weeks ago. And yeah, we we're definitely, definitely putting ourselves as a short-term risk, right? You can swap us for anybody else. The hope is that you don't, because we've earned the right to be your partner. And because this is a services arrangement. What I don't want to be is basically like an managed service provider that you throw a ticket over the wall and we throw it back to you. And then it just becomes very divorced and it's like effectively like, you know, as a service, but only in name.

22:27 What I much rather prefer is to be kind of almost embedded with you and your team, to be an extension of your team. That can scale up and down. And there's some, you know, we have to manage that on our side, the economics of it. But You know, what what what I found in in in my career is that if if you and this is potentially naive, but if you deliver good work and you keep showing up, your customers will reward you with that, like for that nine times out of ten. Sometimes it won't go your way. Other times your customers will go out of business. You know, like things happen, exogenous variables kind of happen and like happen to you. And all you can do is control what you can control. Going back to stoic thinking.

23:08 Yeah. So what y so that's what you can control and that's what you can do. And and Here's where I think the naivety here's where I think the naivety is true. But fundamentally, what I've learned in my 47 years now in multiple businesses, and by no means the expert in business, but what I've found is it is around value. When when you, the business, provide a high value in every deliverable in the relationship that you have with all your clients, whoever they are, you're gonna win. And they're going to be more loyal because it benefits them to be loyal. It's around the value play. Because if it's around, I just want people to just be loyal because I've worked with them for three years, or they like me, or they started with me.

23:57 I have found that is a naive thought and approach. But if you provide an elite level of value that you can take to the marketplace and that wins compared to other people there, you will have reciprocity. Yeah, that's that's what I'm hoping. ~ a couple of other thoughts on that. One is this is like a referral business as well. Like if you if if I'm working with you and then you're at a dinner party with somebody and I mean, I don't know how often data comes up in your conversations, but Bus no business owners talk about their problems together. Business owners hang out with other business yes. Totally. And they and they talk about AR. I mean AI, everybody's talking about it. And you're like, you know, I've been working with Ilan. He's like pretty good. You grab 30 minutes with him. That I I would say like half of our business so far is that kind of referral.

24:41 The other half is my network. And you know, eventually that that'll that'll expand and and grow. But I'm a big believer in like that long game of like you plant seeds, you do good work, you have a good reputation, you're a kind of a stand-up guy. And and it comes around. The challenge is going to be delivering. Service excellent service at at margins, at like a margin that makes sense because services are costly and they're hard to do at scale repeatedly. You know, some companies do it, right? Like McKinsey and the like the bank, like BCG, like they've they they do it for their business and and it and it and it happens in others. And I think from the AI services perspective, like that playbook has not been written yet. I'm I'm happy to be part of writing it.

25:23 And and and again, I think like AI. It it sounds a bit of a cliche and you know, especially as I say it a lot, but AI has changed the rules. Like it it's changed the game in a material way, so that things that weren't possible five years ago or they were like accepted wisdom, like we just know that that's not what what's possible because like we tried it in the t in the early two thousands and then we tried again in the twenty tens. And well, let's try it in t in the twenty twenties with AI. And I I I I suspect that some long held beliefs about how this business could work at scale are gonna be challenged significantly. And I think AI is gonna be a big role player. What do you think? What do you think the things that we believe right now and that you believe right now, what do you think's gonna be challenged significantly?

26:04 Are you curious to be thinking about that as a business owner? Like what do I believe right now that I think can be the foundations of this business that might be dispelled in eighteen months? Yeah, one one immediately a few. One immediate thing that comes to mind is the sh the organizational structure. And I'm open about this on my on my social posts and like with my team, with everybody I talk to, is I'm interested in a in so traditional d kind of delivery ~ models or services firms, you know, there's like a few archetypes. There's like the engineer, there's the account executive, account manager, there's project managers, there's salespeople, there's solutions architects. There's like different folks that that that operate.

26:44 And I'm very interested in building an organization, at least for now, and this is this is up for to be challenged for sure, where you're either you build and deliver or you sell and deliver. Delivery is the one that both but you sell or you build. And that's you know, and I'm not the first to think about that. I know ~ I forgot I f his name escapes me now, but the guy who founded Twitter and then block now. Jack Dorsey. J yes, Jack Dorsey. Very good. Thank you. So you know, he he had this post a few months ago about like you know, the shape of an organization, how everybody's on the edges of this like circle, and then in the middle are the agents and the AI. And I don't know that I'm going there yet, but we're thinking company brain.

27:25 We're thinking only people who can deliver on the outside. We're thinking middle management layer. It gets squeezed. Now I don't want to be doom and gloom and people are but you know, and and when I say middle management layer gets squeezed, it means like at the top is the sellers and the like leaders. And at the and I don't want to say bottom, but like, you know, The other folks are people who deliver. And the the facilitators, ops people, probably at the enterprise still have a big role and and and and might add paradox someday. But for now, ~ I'm trying to build ~ an organization which we don't have those individuals. there's a lot more accountability, again, either for a number from a sales perspective or for your clients from a delivery perspective. And that's kind of it.

28:05 Simplifies the matter. Actually, my customers like it because they get to deal with the engineer that then turns around, does the work and then comes back and has You know, there's nobody, no administrative layer. So anyway, so it's like that's that's one thing that it it just I don't think it could have worked a few years ago. ~ and AI has enabled us to just move faster in every domain. My my leads, the engineers obviously writing code and shipping code, debugging things. When I talk about like risks that services firms take ~ when you're doing pricing, you know, there's like different ways to price. But one of the risks that you think about is like, well, what if we go over budget or under budget? Or or rather, like if something takes longer than we thought. Mm-hmm. Where does that risk live? Well, if it's an hourly model, it lives with the customer. Just keep billing until you're done.

28:50 If it's a fixed price model, you own the risk as the service provider. And you know, different companies have their own perspectives, but hourly, I just don't see how anybody's doing that anymore. Yeah, where do you think the business and and I don't want to interrupt, but this is really something I think that's important for even the operations of your business and where service businesses around AI, how they're going to evolve. Right now, you have two categories of significant expense for your business human capital, engineering, building it out. Or two, right now it seems there's a big argument, and I don't know if it's a myth or not.

29:31 That the AI companies are significantly undercharging for the cost of energy for the AI to be produced. And when the data center stuff gets ramped up, is your cost going to go through the roof from energy and tokens to operate the systems? So is it human capital right now, or you've already dealt with that? And is it going to go to a significant energy expense? Where's the expense lie? Yeah, so right now it is like mostly COGs, like it's it's mostly like the human capital for sure. We are seeing our token spend go up for sure, you know, as we build more, as we scale. And the models, you know, it's so hard.

30:13 I I I discount anybody's prognostications when it comes to this kind of stuff, incl including my own, frankly, because yes, models are getting bigger and more expensive, but then Opus five five point five just came out the other day and it's cheaper than than the previous version. So now you're like, wait, smarter models can be cheaper and Yes, but then you're gonna use more tokens, so absolute cost goes up. It's like there's there's so many factors, and then you and then you weigh in the the open weight models, which are free. They're not really free, but they're you know, let's say much more inexpensive than then so then you think, okay, well now we need model routers, right? And there's open router is a very big company that that does this. You can route models. Okay, but an AI system is not a model, it's a chain of agents doing different things.

30:56 Some of those agents will be You need the most expensive, smartest model. Some of those agents just do very dumb work and you need a cheap model. In fact, some of those agents shouldn't be agents. It should be deterministic software. So there's a whole re-architecture that we that we it's like every six six months, the economics change from underneath us that we have to re-architect the systems, which I think is technically very fascinating. But it makes sometimes I think like what's gonna happen? I don't know, it's a fun, you know, let's have a beer and like think about it. But I also tell my customers I'm not about to tell you what's gonna happen. Let's get to work. Like let's get yeah, let's get to work and do the work and create be a part of what happens. Yes, let's just like start doing. There you go. And then, you know, you have an eye towards where things are going 'cause you don't wanna you know, the one way, two way door is the Jeff Bezos kind of mental model.

31:42 You want to you know, or like reversible or irreversible decisions. If it's a reversible and this I tell my team all the time, if it's a reversible decision or a two way door, don't think about it. Don't ask me, just do it. I trust you. Yeah. Just do it. Well, like w and with AI, it should like I I I I tell them if if you're thinking about a thing, it's gonna take you less than a day, don't like it's not worth asking me. Do it. And you know, if there's no surface area risk, of course. The rever the irreversible, the one way doors, that's we should do so, we should spend some time thinking about that. And from a company perspective, I'd say most of your AI decisions are going to be two way doors. You know, is it the right vendor or not? As long as there's no significant lock in and your data doesn't leave your environment. So there's no like long term risk, just like pick a vend, like great, that vendor. How quickly can they get you something?

32:29 Like that's it. Give me an example of a one way door that makes sense to somebody who could who is gonna gonna listen to this and they're your potential customer. How do I understand what a one way door and the thing that you would build with me would be? So that's that's that's a good question. And I actually think like a lot of one way doors are are are going away, but like traditionally it might have been building versus buying, right? We want to build our own, let's say like a CRM system. This is happening everywhere right now. Everybody's like, I can vibe code a CRM system. I don't need Salesforce in a HubSpot or Atio or Pipe Drive or one of these. The answer might be you're right, but not obviously. It's one of those just because you can doesn't mean you should. So it's a if you were to build a thing.

33:09 The building of the thing takes some time and investment, a fraction of what it used to. But more importantly is the change management across your organization. Get everybody on board, like re-rethink your workflows, redefine things, really orient around this tool that you guys have built, for example. And in particular, if you're building it with an external agency, that's another risk because if that agency goes or if you don't like them and you want to swap them out. Well, some other vendors now gonna have to read all this code and understand it and they'll probably wanna do it themselves anyway. So I'd say that's not a one way door, but that's a big decision versus a buy. And and I often tell my customers, like, buy something cheap off the shelf and let's just see what happens.

33:50 And it's not gonna be perfect, but that's okay, because the more you live in the world of pain, when when we talked again in six months, you'll have a very clear, very precise view of what you want to fix. As opposed to some amorphous, I want better reporting. Like Okay, I'm with you, but like what specifically? Where are the you know, so live in the pain first. So I would say buy, you know, versus build or spend a lot of time thinking about that. ~ so that's like a big decision. I don't know how one way it is. And then the other one I I would say this is not a one way door, but this is ~ you know, there's a lot of companies I'm finding, frankly, I was a little bit surprised, but this is maybe my naivete. They don't have they're they're living out of SaaS tools. So they have a Salesforce, they have Net Suite.

34:33 For their ERP. They have, you know, some ads manager where they're running marketing ads online. They got some other third party data. And they're happy. They're like, this is good. I'm paying these SaaS tools. And they, you know, and every now and then they they want to do something that they can't in those tools. So they bring somebody like me, we get them a cloud and a warehouse, we bring all the data together, we combine it, and now it's like, it's cool. I've never seen this data before because like the UI wouldn't let me get behind it. And we have the conversation of Well, you never you never owned your data. Like, well, what do you mean I never it's my data. I I input it and this is my specific instance of a CRM. Like, yeah, you're renting it.

35:15 Like, what what do you mean I'm rent? And then I explain, can you get it all out? Can you do what you want with it? And they're like, Well, I can download to CSV what they allow me to download. I'm like, does that feel like ownership or are you renting? And they're renting. And when people have this light bulb moment that they're like, You're right, then they think, Well, does that matter? Maybe I'm okay renting. I say, maybe you are. How do you think about security? How do you think about competitive differentiation? How do you think about like using AI on top of these tools if they don't let the data leave because data's their moat? Your your data is their moat. Just kind of a weird, but that's is the case. And and so like going back to one way, two way doors, if your entire business is built on SaaS, I mean, the time to like take control of your business and your data is now.

36:01 the one way door there is like not doing it. So I'm thinking about this from all your so here's it's funny. See, when you deal with you brilliant guys, when you said you were surprised when you got into organizations and they're operating on multiple SaaS programs, if you would have talked to an operator like me who doesn't have the trained, educated intelligence the way that your data science brain does, I would have instantly said, That's ninety percent of companies, Ilan. That's my opportunity. It is. That's what I'm saying. I'm so what I'm telling you is when you said it I'm like, What do mean? That's every company.

36:44 Yeah, but here's that. That's how that's how Oracle and Larry Ellison are the most valued that's how, you know, that's how Salesforce became that's how that's it. This is what companies are built on. Right. And these tools are great, by the way. You know, like I'm not saying like get off the like use the tool. Yep. But just be clear, you know, unless you have a strategy to to to get your data out, you're renting it. That's it. Yeah, we're going through it right now. ~ A back office in my real estate company, back office, ~ transaction management, e-signature, one of the huge companies. But we are actually, we do have the opportunity, but we've had our staff in-house, we've been downloading our historical data for a week and a half to plug in, to do some things. But what but thankfully we had access and we could get it, right? Because as you said, now's the time.

37:33 Right? If you weren't doing it, now's the time, right? Here's what's funny, which which I wonder. I don't know what this is gonna mean for your future companies and what you're gonna build. What my prediction is of AI in business is that I don't know because the 47 year old me hopefully has another like 25 years in business. So I'm here. Right. I was I love where I was at. I was at the beginning in high school. It's when AOL and it was dial up. And so so I've gotten to see this entire trajectory. But now in business, here's what I'm here's where I'm what I'm going through in my brain. I've committed myself from now through the rest of the year.

38:16 It's funny that we're sitting here talking on my podcast, you're building your business. I said I'm committing the rest of my year. To become as well versed as I can in what AI now means for entrepreneurs and business and me. Because I think the future business owners are gonna have to look more like you than me. Yeah. So a hundred percent. Like the next generation. Yeah. And I'm and and I mean I see this kind of in real time. You think about all like the new crop of businesses, how fast they're growing. Now I think some of that's funny money, so I don't really I don't really take take too much stock in it.

38:58 But if you think about, you know, today's executive, today's CIO or CTO or CEO and all the rest of it, versus twenty years ago, they're they're a slightly different breed and I think that evolution's gonna happen faster and faster. It's gonna be compressed. So now I can talk to many business owners, you know, it's of slightly larger organizations. I can I can say the word API in the context of systems. Yep. They understand. I mean, they don't know what and but they right. That's how systems talk to each other. Yep. That's what I've been interacting with with accessing our data over the past week and a half, two weeks. Right. In API. Can you imagine fifteen, twenty years ago talking to a gray haired executive about API? They'd like what? I mean it existed.

39:40 It was it was still it was it was very fresh fifteen years ago, but it existed It wasn't necessary. It was nuanced. Right. But I actually believe now, man, is the threat to current business owners right now, and maybe this is the the the hole or the the that you're gonna plug is that They're gonna have to know how to interact and operate AI systems, man. Yes, the the yes they will for sure. Say no, push back though. Push back on what you're seeing. No, no. so yes, and we all have to figure this stuff out. Like th there are not people today who are I mean, the experts today are like they were not experts a few years ago.

40:24 In terms of using the systems. Sure, there's like you know, researchers who've been doing this for a while, but this stuff is ch is ever changing. A task that I was doing manually in in Claude, Claude Co manually, but using you know Claude code to write code to do a thing, is now Right in the tool. I deleted my entire code base. AI deleted and AI made it. And now I just do it right in the tool. Great. I'm kind of an expert, but I didn't know that was even a thing until I was I was in the car with my wife and she's like, can you make an agent an automation that looks at when schools have tours opening so we get notified and I like how Yeah, I can't. How long was this was that eighteen months ago? That was that was two days ago.

41:06 So can you and I and I said, Yeah, it'll take me a couple hours. It'll take me like half a week and you know, I could write a a script that does it and it traverses and totally doable. I just need to find a couple hours. On the car ride home from wherever we were, she was like, Let me just ask ChatGPT on the phone. And I was like, Yeah, but it can't it can't connect to things and then it's gotta do and then it's gotta run on the cloud. Like, I don't I don't think it can do that out of the out of the box. And then she did. And then before we got it is really crazy. Before we got home, I had an email being like, here are the schools in our like district and here are the tour dates. And I'm gonna be listening every week and I'm gonna send you an update. And I was like, Cool, I'm done.

41:48 And and and I'm somebody who knows what they're talking about, like a little bit. I can only imagine in twelve months, people who are really just trying to catch up. Yes, they have to become so c kinda to to to your to your point, they have to know how to use this stuff. But there isn't a like box to check, like now I'm an expert or now I'm safe. It's changing and we're all changing with it. So you can take solace in knowing that there are no experts, which which to the very beginning of the conversation I think is more important than ever. Not the EQ stuff, which is very important, but that first principles thinking. If you can think in if you can think in a principled fashion and you could ~ you can then kind of build up your your mental models around these things.

42:32 So you understand there, 'cause we're going we're entering a world of design, right? So when you say they gotta use agentic systems, they need to know how to design like how they're designed and architected in such a way so they can prompt them and they understand data moves from here to there and they think about schedules and you know, this systems thinking. Do you think your value at a high elite level could be as your model evolves as a consultant? Yeah, well yeah. So so For a high dollar consultant, the the same way that companies now have a a lawyer on retainer. They're accountant, it's a retainer. Totally. Right. So our business model is once we deploy it, we are on a retainer then. We manage it for you in a much s reduced capacity 'cause you don't need as much once we've built it.

43:19 And we're it's kind of like a retainer, you know, bucket of hours per month or something, you know, kind of depends on what the customer wants. But that's absolutely where we're going. Cause you need to have the business owner be able to drive the AI car, don't you? Some we can automate, but don't they they need to be able to drive it? Someone there does. So y yes and no. And the the yes is like yes they need to like and I believe like they should own it and like drive it. The the no part is it depends on the the executives that I deal with. Yeah. They they like it, but they want somebody else to do it. You know, so if I enable them in cloud, like, hey, you can ask any question to your heart's content, they're like, Cool, cool. Hey, can you give me a report that does that shows me revenue by week?

44:01 I'm like, Well, you can do that in cloud they're like, I know I can. Can you do it for me? Like Okay, I guess I'll get you I'll get you a data team, you know. So it's kind of funny. I have been noticing ~ that that folks want the ability, but that doesn't especially and this is maybe an executive thing because they're you know unique in in some way. They want the ability broadly, but they kinda want somebody else to do it. Like somebody w to do the thing. ~ and and and and that's been an interesting space to navigate. It's like it's not enough to enable someone. You kind of sometimes have to bring them You know, it's a horse to water type of thing. Like you really gotta bring him over. But that was my argument of ten minutes ago. The future executive's gonna need to.

44:44 You think they're gonna ni like they're not gonna have a chief? No, you're you you're you can y I think the essential part is to have you and someone here build the thing for us, educate our brain, inform us, consult us how to use it, and then it's gonna be essential that the executives at the C level are using it to operate and run their organizations. That's what I believe. And and and to the point about like the changing executive, I do think this mindset of can somebody else do it, I mean there's something innate to that in all of us, right? That's why SAS is popular. But I also do think that the next generation of executive is gonna have like grown up doing it and being able to. And I I suppose one version of it is in ten, fifteen years, when the next generation has taken over, they're gonna do it themselves.

45:30 They're not gonna need the apparatus that could But your advis your advisor and consultancy business that we call SAS now will be called something else because it'll be so highly valued. Even if an organization pays you a a half million dollars a year, pays you 500 grand a year, that would have used to have whatever the benefit is in the consultancy you're giving them, that probably would have used to cost them five or ten million dollars. It's going to be so valuable. And I do think that's where the margins in your organization can go through the roof. I think they could actually believe. I it seems to me that they could almost be reversed. Where if SaaS was running on a 70 to 90% margin, I think you can operate this in the future running on like 99% margin.

46:12 I think that I that's what I believe. And you're in the business, but that's what I was talking to these guys in Miami. It's like, you all, I'm not in this business, but I see what could happen here and what this could be because I'm a business owner that sees the value. If you give me access to something for a half million dollars a year. That previously maybe my organization didn't have the revenue and resources to pay for. If I really am better than my competitors, that's gonna unlock in twelve or twenty-four or thirty-six months my exponential growth. And all you got to do is pay an Ilan five hundred thousand dollars. I I think it's a good deal. No, but but but there will even be small, small companies that they might pay you fifty grand, right? And then you put on some monthly thing.

46:54 There'll be some Right, exactly. And that's that's kind of the plays, you know, we want to do we want to meet the customers where they are because this is a very elastic offering. We don't need to give you the whole world. Sometimes you don't need it. So we can and because of A ~ it it doesn't enable us to scale the offering. So it's not just like you get like tier one, two, three. It's like it's customized to you. And by the way, in twelve months, when you're growing, we'll grow with you and your needs will change. And that's that whole partnership thing. Yeah. So give us man, we could sit here and talk about this forever. I just I think what what our listeners and the world will appreciate about this is just just practical, honest, transparent conversation around the subject of AI and what's happening.

47:34 Right. Because what without because even though you have a business and you're building it and you're branding and you're trying to raise awareness, that's how you sell, right? Podcasts are a platform to do that now to gain exposure where you might have had to use to get on, you know, Fox business or CNN business. You don't have to do that anymore. So But I think they'll appreciate the conversation. However, bring this back down to like the one or two thousand foot level on what your pitch, even though you're a guy who knows how to sell without having to market, what's the pitch for the average business to utilize these services and what you're building and how you want to help them? Pitch is if if you find yourself stuck in spreadsheets or repeated manual processes doing things like ret reconciling numbers or stitching things together for the purpose of a report, for the purpose of understanding your business and doing analytics, for the purpose of using an AI that somebody told you you should use or you believe you should use.

48:38 And it's a little harder than you think it should be. The problem is not you. It's something with your systems and your data and your processes. And those are the types of organizations that we like to help. And that's how we like to help them. What's it look like if somebody says, man, this is the guy that I need to help? What's it look like for their onboarding and time commitment? And you're not gonna give an exact cost, but businesses are all different. What's that look like when someone says, I need this service? What's that look like for me, the business owner, then? Yeah, so kind of we'll do a discovery. We kind of understand your needs, your business, understand like what are your very specific pain point. That's a couple weeks long. We'll try to make it fast. then we build. We like go immediately to building.

49:20 I much prefer to build and get feedback from you. So we build immediately, ~ a couple iteration cycles, we build a thing, you tell us, give us some feedback, we keep building together, and then it's done. And then it's deployed, and then we manage it for you. Two, three, four month project. Kind of depends on the size and scale, but I don't I don't sell 12 month, 18 months. I I heard of one vendor selling a 24 month implement. I was like, how could you do that with the straight ~ in in in in my people's vernacular that's called chutzpah? And that's like you know, the chutzpah that somebody has to sell you a 24 month project. Like I I wish I had that confidence. No, we we're talking two, three, four months, and we don't just like walk away. We're there, we're your partner, and ~ that's what we're doing.

50:04 Do you think I think these things you're building now and everyone else, they're gonna become legacy systems faster than anything ever has. So do you feel that these will be able to be not just built on? Because I go back to like 30 years ago, like phone systems. Anyone who knows phone systems, once you bought into a phone system, my gosh, to make a change. So you're smiling, right? You really couldn't improve it that much. You couldn't. And if you did, it was expensive. It took forever. Does it seem like the things you're building cannot just stack on stop on top of the same legacy thing?

50:45 It truly can evolve and grow as AI grows. Is It yeah, it absolutely can evolve and grow because we're writing code. Like it's low level code. We're not using a third party tool. We're not the whim of anybody. It is it's it's it's low level. And with AI, I mean, AI helped to write it and AI AI will help to augment it. and it's infinitely portable. so there's no lock in, both like from us as a vendor, but also, you know, this is a little too detailed. ~ but like the way data is stored, there are formats that don't like move. You know, there's like CSV and text files. There are certain formats of data stored in a way that, well, you got to use a vendor that plays nicely with this format. Otherwise, you know, you're kind of SOL.

51:25 so we're also building in a format open source, open source format that is portable. So we're we're really lean leaning into this. We want to empower you the business. And and in like up and down the stack. So so yes, I think what we're building is gonna is gonna evolve. And the key is that in the past, you can measure legacy from like the time of implementation or deployment. To where you are today, the day after deployment, and then every day after, the legacy doesn't change, but your business evolves. And eventually you look back, you realize, we're like very far away from where we were. The new systems, they're going to be evolving with your business. They don't want to be ahead of your business because that's a little too volatile.

52:06 But if they can be just a little bit behind, weeks, months, quarters behind, not years behind your business, then I think you're in a good place. And that's how I would architect these things. Does there need to be someone designated inside the organization that you all work with? If it's a small organization, is it the business owner, if it's a little bit larger, is it some One on up. Yeah, yeah. So every every project, I mean, one of the biggest failure modes is that there's nobody internally to like unblock doors, sign the documents, get us the access, can can break a tie when there's two different people arguing over a thing. Like there's only so much we can do as as external. So so there has to be like an executive sponsor. And if they're not day to day on the project, there should be somebody who's day to day on the project.

52:48 Even if we're the ones managing everything, we need to have like very quick feedback. I I never want to be too far away or f too far behind where again, where the business is. right, hey, you could give us all this time, but you got to take your brilliant brain and go build things and take data and manipulate it and go help us build businesses. But I I I appreciate it, man. It's so refreshing to talk to someone how we started out that's in this space, but that also has the ability to connect and feel normal, right? And so it's not it's not too overwhelming. So we we appreciate that. But tell us the place that as you're building your your business, but it's Paradox Machines. That that's the company that you're building. What's the one place? Is it following you on LinkedIn?

53:28 Is it going to a website? Where do where do we go to engage in yeah? Yeah, I would say follow me on LinkedIn and from there you could find the website and the company page and everything. But LinkedIn's a good place. I'm there, you know, a couple of days a week, doing my thing, you know, a little bit unfiltered, but hopefully that's okay. And yeah, that's that's the best the the best place. Okay. And if we Googled Paradox Machines or if we talked to chat or if we talked to Claude, what's gonna come up? Do you know? That's a great question. If you googled it, hopefully the company, because that's kind of a unique naming. I th I think I think we'd come up. I think Paradox well, so there is a Paradox machine is is like a thing from was from ~ from something from the sixties or seventies, some sci fi. So it is a thing that exists.

54:12 So that might come up. But but hopefully hopefully we'll be near the top. What's the meaning of the behind the name of the business? paradox well I mean we're definitely gonna run out of time. But Okay, I I I have another five minutes. If you do, yeah. Yeah. In short, paradox, I I'm you know, I'm I like philosophy. I like to kind of think about these kind of whimsical ideas. I've always enjoyed paradoxes in every manner, actually. It was like fascinating to me. And then, you know, very famous paradox, Xeno's paradox, you know, he was like very popular with this kind of stuff back in the Greek days. But ~ and there's a big paradox with AI and data. Simply it's everywhere. It's it's a signal to noise. It's everywhere and yet it's it's elusive.

54:55 You know, AI is gonna create jobs, but it's also taking away jobs. AI is the future, and also it kind of is dumb, you know, in some respects. And these paradoxes, these multiple truths that we have to hold and and they're competing. And just like untangling this paradox, ~ I find that's a big part of what I do is, you know, provide, you know, guidance and partnership. And yes, we build systems that work, but we also, you know, as we said, it's a relationship business. And if I can help untangle the paradox that is, How do we get this tech to work? How do get our data to be useful and be quote unquote data driven? That that's kind of a cool place to be. Okay, guys. He's a great dude when it comes to AI. So Ilan is spelled different. It's I L A N ~ A N.

55:38 We just had to correct we had to correct that so they weren't going to the other the other. Appreciate you, man. Get back to work and helping all us business owners at this level, man, because we need it. So thank you. I appreciate your c the conversation, Mark. This was this was great.