← The AI in Business Podcast
The AI in Business Podcast
AI for Driving Human Reasoning over Reasoning Models in the Enterprise – with Chris Busch and Yancey Sanford of MSTRO
00:46:06
Today’s guests are Chris Busch, Founder and CEO of MSTRO, and Yancey Sanford, Founder and Chief of Information and Research at MSTRO. MSTRO develops an advanced AI platform that combines data and creativity to optimize human performance. While AI...
Loading summary
Transcript28 lines
- [00:00]
Chris Bush
Foreign.
- [00:07]
Matthew DeMello
Welcome everyone to the AI in Business podcast. I'm Matthew DeMello, editorial director here at Emerge AI Research. Today's guests are Chris Bush, founder and CEO of Maestro, and Yancy Sanford, founder and Chief Information and Research Officer at Maestro. Maestro develops an advanced AI platform that combines data and creativity to optimize human performance. While AI reasoning models promise efficiency, are they the right approach for businesses? In this episode, Chris and Yancy join us to challenge the prevailing wisdom behind AI driven automation. Throughout our conversation, they argue that enterprises may be too quick to offload decision making to AI when human intelligence remains the most adaptable and IRreplaceable asset. With AI adoption accelerating, business leaders face critical decisions about how to implement these technologies without sacrificing innovation or adaptability. Chris and Yancy lay out a compelling case for keeping humans in the driver's seat and ensuring AI serves as an enabler, not a replacement for enterprise intelligence. Today's episode is sponsored by Maestro, and without further ado, here's our conversation. Chris and Yancy, thanks so much for being with us on the program this week.
- [01:32]
Chris Bush
Thank you so much, Matthew. It's a pleasure to meet you.
- [01:34]
Yancy Sanford
Yeah, thanks, Matt. Thanks for having me.
- [01:37]
Matthew DeMello
Absolutely. I'm really glad we're doing this podcast because I think this is an underlying conversation. I don't think we've had time on the show to really pull apart just yet. And with the dynamics of the market, a problem and a challenge across industries that a lot of executives come on the show and talk about is just the speed of these technologies is outpacing our methods and our standard operating procedures for keep up with that technology. And that's going to present more challenges into the future, not least of which we're going to dissect on the show today. And that's principally that a lot of AI markets across industries seem to be in this race to building reasoning models, making AI systems more expensive and computationally intensive all at the same time. There's this growing fear that not only might AI replace human jobs, we've kind of covered that on the show just in terms of we're going to see a lot more jobs displaced, but moreover that we're going to see a lot of these systems maybe in an inappropriate fashion, but replacing human decision making in a way that might ultimately hurt businesses in the long run, particularly as they focus on AI driven productivity. But wondering if we can kind of take a step back here, magnify the challenges we're seeing in driving reasoning models to automate maybe things that humans should be a Little bit more in the driver's seat for. But are we thinking about AI's role, role in the right way? I guess is the bigger question, are reasoning models truly the future of AI or is there a better approach? And we'll start with Chris, of course.
- [03:20]
Chris Bush
Thank you again for having us on. I think if you take a step back and you look at the market like this and you look at how humans really like are in the world, it's back to where we're really tool builders at the end of the day, right? And when you think of humans as a tool builder, you look at the, the technology and the direction that it's going and, and really the, the energy efficiency and the, the ability for, for, for humans to make decisions where we're suddenly thinking that AI could do that for us, Right? And I think, I think that the problem with that is, is, is right in the root of why everyone thinks AI systems are wrong or incorrect or don't really provide you the right answers. And that comes down to the basic human instincts of what makes us tribal or human. And it has to do with the fact of really understanding if you believe it or trust it because you didn't really train it or it's not really part of you to trust. And that's where I think it comes down to where Maestro. And where we've looked at is, is we really brought it back to the roots of what humans are best at and we're best at building tools to make us more efficient, right? And everyone else is really looking at the market of making productivity tools to replace the decision making of human. And that's back to the reasoning, model, logic that you're getting that you were talking about before Matthew and I truly, I throw out this kind of thought process. When you think about it, if you think about like the times where like each of us, we sit there and we have complex problems and we're trying to solve the issues that looking at and we just can't find it even through the data analytics and everything else that we truly like and see with everything. And the reason being is we might just not see the right angle, right? So we might step away, we walk away and we think about this in an aspect of like, okay, but subconsciously you're still thinking about it and you might go for a walk or you might be hanging out with your family or for me, like, I watch stupid space movies and my wife hates me for it. But what it comes down to is suddenly as I'm watching it, it's like that those dots connect. It's that aha moment that I have and that's what Maestro does and that's what human, our ability to connect dots is what makes us special. Right? And that moment is exactly where you get innovation from. And we at my show believe that if you provide a tool that allows the human to reason through the data and provide them the ability to have that aha moment. Why do you need reasoning models at that point? The human is the reasoning. Because the human from that process of exploring and discovering is allowing the ability for us to now truly make logic of what you can't teach a system and a reasoning model to do. At the end of the day, when you look at the market, basically what you're talking about is training a model on data. Well, data doesn't have like human intuition, experience or that aha moment I just explained to you because we can't see that, that angle.
- [06:38]
Yancy Sanford
Funny, because if you ever seen those ads back in the 19th century where the hot air balloon came out, right? And everybody was predicting that everyone's going to have a hot air balloon, their own personal hot air balloon. So as human beings we have a tendency to look linear at like predicting the future and how things are going to roll out. And it's when you take a step out and you take that trajectory down a way that they're not really looking. Right now the AI industry is looking at just like let's throw more CPU or GPU into the processing, let's make these billion dollar data centers and just make it bigger, bigger, bigger, bigger, bigger, that in the end as humans as we progress is always tends to fail. So what we are doing is we're taking, we're seeing that direction that you know, the bigs are doing and we're saying, well, you know, if they're trying to make this AGI and they're trying to make it so big and they need all this processing stuff, maybe we don't take the human out of it and a human can help dictate how that grows and how it goes. Take away from linear thinking, spatially thinking about how the human integrates with AI.
- [07:58]
Matthew DeMello
Just even pulling from not only what you said, Yancey, but what Chris was mentioning in terms of there's a short term roi, a tactile roi and there's a larger vision here just in terms of the role that humans have as we go forward. You mentioned AGI, that's a little bit more into the future. I know our head of research, Daniel Fagella, loves to talk about that stuff on a podcast he's got called the Trajectory. I'll make a quick plug for him while we're here. But the point being, you know, just in terms of what you're saying with these reasoning models is that we might be a little bit too fast just in terms of where we think the business value might be for really driving these systems into place, where human beings are right now, versus there might be more roi, at least in thinking of human beings as still being responsible in these roles, especially where there's a lot of concrete decisions tied to the business goals here that maybe we can magnify. I'm wondering, just in terms of a lot of the times, I take it, that maybe these larger firms deploying these reasoning models, really putting all their money down for AGI, they might not see their challenges or they might not see the efficiency, or as we used to say in global taxes, you might not see that money that you're leaving on the table. But what do you think are some of the disadvantages, at least in driving the reasoning models? What are some. They might think that they're driving greater automation and your tool does traffic a lot, at least in automation, in being able to have human beings determine what gets automated. But I'm wondering, as we're deploying reasoning models, what are you seeing as sort of the business value that leaders are leaving on the table by being so quick to drive with reasoning models in these solutions? Chris, we'll start with you.
- [09:53]
Chris Bush
Yeah, so. So if you take it like, think of it like this, and everyone's in this race for productivity and they think that using AI, we're going to get automation, right, and we're going to get RPA glorified, that's really what it comes down to. But, but if you step back and you really look at what AI is really doing for us, right, Humans, at the end of the day, and back to the kind of the original question that we were really talking about here with, with these reasoning models is, is humans at the end of the day, we're great at processing, but we're slow at it, right? And if you can speed up the ability for the human to process what they're, what they're trying to look at or see or trying to figure out, then you're really playing to the aspect of saving the process of how they think through it. And that's really what reasoning models are at the end of the day. It's how they reason with the data that they're trained on. Well, here's the problem. With that they're only going to be as good as as the last data set that you trained it on or the last aspect of where your data is. But tomorrow, and data changes every second, it's outdated already. That's why everyone's in this boot reload and they're always trying to continuously train models and think that they can get a smarter reasoning model to help automate a task. But if you look at AI that's scratching the surface, that's looking at it from almost a service desk logic of the world. What we're talking about is truly capturing the human's ability to process our uniqueness, what makes each of us unique from the standpoint of how we see the data or how we see the problem or how we solve things in our micro aspect of our organization that you can't capture at a macro level that you have to capture at the moment of how the individual solves the problem, their intuition, their experience, or even down to the the aha moment of how they connected a dot that no one can see. Most organizations think that process is going to drive automation. And the reality is that that might be the success they think it is. But then they get lost in the, in the realm of what the process is and think that's the product. What we're saying is, is, is if you look at Maestro, we're capturing the content, the art, the actual what makes each of your your A players special. And if you focus in on that now you suddenly have a system that can evolve with your business, your dialect of your business. Because what you're capturing is truly the knowledge repository of how each individual does their work right at the end of the day. And that is what makes businesses grow versus just stay alive. Everyone is in a race right now for a productivity tool which is the optimization. But that just keeps you alive during a time right now where you see layoffs happening and you see downturn in sales, everything. And frankly, sure you can survive, but this is the time that the competitive edge is going to give you the ability to grow. And if you can spark that innovation all the way down to every single micro aspect of your employees base from the end point to how it impacts the business and understand the art of how they do their job. So special. And now you're talking about elevating human intelligence versus just automating simple tasks.
- [13:14]
Yancy Sanford
I'll just kind of simply, I guess summarize it like this, is that a human with AI is always going to be greater than AI one. There's your I guess Competitive or roi, we feel our direction is better than current market situation.
- [13:35]
Chris Bush
But even to elaborate on what is he saying there? If you look at it like this, everyone's throwing out, oh, oh, chat GTP or deep SEQ or any of these large language models can be like our scientists or mathematics or take these tests and everything else that's great, but where's the creativity? Where's the ability for the business applicability of it? Right, Whatever. What I'm getting at is cool, that's pure science. But. But at the end of the day, it's got to be applied business. And if you can take a test, that's great. But if you look at most of the kids out there or students who get great grades and do amazing at school, that doesn't mean they're gonna be great at business or about what they're gonna do with their job. It's academics versus business at the end of the day. And I think what we've gotten lost down is a paradigm where we believe that, that by teaching AI to reason for us is going to make us smarter. But in essence, you're starting to see the fallout already of where it's actually removing the humans creativity and inability to truly keep the world vibrant versus turning the world into a gray dark area of the same over and over. And I think that is also showing why we're also seeing how it's replacing humans in the workforce and everything else. Because the problem of focusing on that aspect is cool technology, but it's not going to connect human intelligence.
- [15:05]
Matthew DeMello
Absolutely. And I mean, I've got a Twitter feed, social media feeds full of creative people and musicians who are extremely sympathetic to that argument and are listening for more business leaders like you to base their technology on just that. All that being said, hey, for the of this show, it's really more about, you know, it doesn't matter if it's reasoning models or human beings. The real question is business value. And I want to go back to something you had said Chris in there, just in terms of the, you know, the difference that you think a lot of folks get wrong in terms of, you know, this big drive towards reasoning models, which is human beings are perfectly good at decision making. They're actually underrated in this respect. You were mentioning how, you know, we can train an LLM to be a scientist to pass the lsat, but where these are enterprises, they're, you know, up until fairly recently, maybe the pandemic, they were places with single locations, you know, four walls, a ceiling and a floor. And it Was all that business context that only a human being, at least at this point of technology, can experience that it was that it was their institutional knowledge that businesses are already driving right now with training processes they don't need to. To your point, take an LLM, train it on their standard OPER operating procedures. Heck, they're even circumstantial employee seminars, the sexual harassment webinars, all that stuff. It's much easier to just take the large language models you have made of flesh and bone known as human beings that have already processed those trainings, already looked you in the eye as an employee and said, yes, I'll accept the job because I did a successful interview. Those folks have better institutional knowledge. And you run tremendous risk with trying to catch an LLM up, up to that level of understanding. You run tremendous risk no matter really what kind of platform you're trying to drive, that amount of human understanding. So that seems to me to be kind of the crux of your argument for keeping human beings in that driver's seat. Speaking of driver's seats, this is something, or at least a comparison that we make all the time. When we have auto manufacturers on the show, they talk about a very special kind of uncanny valley. That's a popular term. It goes back to the beginnings of technology where a human being will reject a human looking robot because it looks too much like a human, but also not enough, which gives them that creepy feeling. There's an uncanny valley emerging in auto manufacturers and how they're developing self driving cars in that they'll build so many features into that automatic system that the human driver will start to check out in unsafe ways. They make inappropriate assumptions about what the system is capable of handling on its own and then just start not driving. And the technology's not there yet. And that's a very, very dangerous gap. It seems to me what you guys are kind of arguing is not only is that gap there for business decision making, even when you're articulating a tool like yours that lets human beings determine how processes are automating under it, we should never get to the point of the equivalent of self driving cars. In this respect, humans will always be better. You will never be able to train a model on every face of your business. And even if you try, that carries so much risk in terms of the trust, Chris, you were talking about before that maybe we need to think about this as keeping human beings in the driver's seat, never crossing that uncanny valley and thinking about how do we automate truly the processes underneath them. That they determine are deterministic are just so matter of fact that a system that has never been through company training, that, you know, even if it takes some form of some kind of AI agent, that it should only stick to kind of, you know, the low order, as I was saying, highly deterministic tasks, highly rule based tasks and keep the decision making in their territory. Do I have that right? Just in terms of, you know, understanding And I'm wondering, you were talking about that, that difference between human decision making and processing. How do we put the processing in their hands and what does that system look like?
- [19:31]
Chris Bush
So look, what you just explained is exactly why the technology today that's being pushed down on us, basically what I would call from one of my developers through at Abby is like the AI overlords, right? But really what I'm getting down to is that's not scalable and we all know it. If you played with these systems, you've worked with them, you're constantly scaling up in cost, price, you're everyday employees not happy with it. Even if they build an agent, even if they build whatever you want to call the buzzword a day, chatbots, agents, whatever it is, right at the end of the day they're just if and statements. Let's be honest, everyone understands it, it's just glorified RPA and that even though it's doing a great job with the little tiny tasks of the day, it's not going to really move your business. It might save you a little money here and there and so forth and I'm sure. But you look at the ROI on that, none of them are profitable. And frankly, I guarantee you're bleeding right now with how much you're spending on trying to maintain them and keep them up from running, right? But I think that is the paradigm here that we're really talking about. And if you look at that, it's really because it's based on a business model, not our business model, not the way you make money. It's based on the AI overlords business model, how they want to process and charging its data. At the end of the day, they want to have the data, see the data, process the data, eat your cake, habit too kind of logic, right? And I think we're all basically tired of that if you really think about it. And I think it's an outdated model that they don't know how to else change because they can't pivot. So where we sit is maestro, we thought outside that box and the way we thought outside this box is if you look at it from a standpoint of what the three things humans do every day, right, we explore data, we discover insights and we create knowledge. That's why every specialty database, every system, anything out there that exists today with a GUI interface or interface is for. It's because at that point, at the micro level, you are creating the intelligence to how it relates to you from that data. And the way that we looked at it is if you serve that up from a user centric standpoint, meaning that the interface molds to what the user wants and sees. Now you don't need any other system at that point. What we're truly talking about is truly an intelligent system that's allowing for an individual to really flow through the data and understand it from a standpoint how they want to. But they don't need to have a tactical system like we have today that is built for a single purpose of a use of what we think it needs to be used for. But the moment time changes, right, it's now outdated. Now we're pulling all the information out and now we're, we're calculating in Excel, we're doing in this or that or whatever, but, but, and then we can push it back in. And, and that's just not a system that's optimized to what users need at this, at this day and age. And, and that's really where you look at chatbots. It's the same thing. You're talking to it, you get information, you pull it out and you make sense of it because your reasoning, right? And then you're putting it back into to fine tune it some more. But by doing it that way, you're also prompting. And that brings us to the reason why they're not scalable systems. Because why do you need to prompt or have, have to learn how to prompt when you already can reason through the data better than it can? You're basically prompting because you're trying to get the system to reason for how you want it to see it at the end of the day. So those problems are how Maestro has been built to now solve basically from a mass market product standpoint, the ability for the user to reason to the data faster. And if we can do that, we save the process. The human becomes the reasoning models in the system. And now the system doesn't have parameters. It can mold with how humans learn, how humans understand, how they adapt, how they, how they evolve through their experience and intuition and gut feelings that a system can't be taught from a data endpoint. And that's where I think everyone is in a race for something that is actually not going to solve their business problems because they're eliminating the human element of what makes their business successful from the art and the content that those humans create.
- [24:07]
Matthew DeMello
Lots to pull apart from there. I want to zoom in in terms of what you see as the solution, especially on this processing side. You mentioned a lot in that last answer, Chris, about even when you're trying to deal with an LLM or leverage a solution using a reasoning model, ultimately what humans only end up asking of these systems is to the data in a way that's easier for them to understand and ultimately make a decision on what's your ideal vision of how that should look and how should technology be bringing what's most important to the user?
- [24:42]
Chris Bush
I touched on a little bit before, but if you look at it from those three elements of what we do every day, if a system can deliver you the data within the context of what you need it to be, to be able to, to connect those aha moments that I described before too, there's truly understanding the ability to speed up the human's processing or the ability for the human to connect dots faster. Because at that point you're able to now suddenly provide us the ability to be smarter in essence of what we're trying to understand. And that's why, for example, I explained if you couldn't solve that problem, you walked away. And when you walked away, it took you you longer to process to finally connect those dots or get that, that inspiration that you weren't, weren't seeing in the moment. So that's the crux of what makes humans special at the end. That's why we're apex predators. That's why we've been successful. We've built tools. If you look at the condor, I think I've heard this before where it's the fastest animal on earth. But if you give a human a bicycle now, we are right. So, so let's think about AI in the way of how we build tools that make humans better, not replace the human. And I think the race to go after reasoning models is exactly the problem of why it's not being truly successful or even showing ROI to any of these organizations. Instead it's putting a bandaid on a problem and we're thinking we're getting success and now we're laying off all these employees. But two years down the road from now, when your knowledge repository hasn't evolved and you have no more innovation, I don't know how much Longer optimization is going to work for you if you can't spark innovation, which is the human. And that's truly where you have to have a system that captures the human to be the reasoning in it versus having a system that creates glorified rpa.
- [26:43]
Matthew DeMello
This is a lot of big vision ideation and really looking at where we're going long term, of course there are short term wins and just in terms of like the incremental business value. Wondering if we can touch on that a little bit for what, you know, business leaders might see. Once we start implementing solutions along these lines, thinking more a little bit about human processing being the weakness, not the human decision making part, what immediate wins do you think we can see? We know it's not all about that, but that always makes a better case for business leaders to win executive buy in if they can show those.
- [27:21]
Chris Bush
Think of it like this, like and this is something we learned from this standpoint and I'm sure everyone understands this is like everyone has their specialty databases or specialty app and like you better not touch it, right? That's not what Maestro does. And that's exactly where you have to understand that that old tech stack of yesterday that you're dealing with never goes anywhere, stays there. Maestro is a tool that just lays across it. And the beautiful part about that is our business model isn't data. We're a federated system, meaning that we deploy it, you guys provision it to your environment, and by doing it that way, it adheres to your security protocols, your guidelines, your data quality, your protocols that's unique to your industry that you've invested money into, stays in place. Maestro provisions by you to it, which allows for you to almost deploy like a GitHub logic if you think of it in that way. By doing it that way, you've now provided the ability for Maestro to basically become that unified platform of data that has the observability, the traceability, the data quality, the data governance, everything that everyone thinks is a siloed solution that they're bolting onto their old environments. That's the mistake because the moment you start doing that, you now don't have a central control system of truly one source of truth. And that vision is where Maestro came into play, where we built it from a standpoint of allowing the technical folks to still do their technical job and be able to, to maintain their CMMIs, their ISOs and even down to their security protocols, their active directories, everything. From that standpoint, then we deploy an app and the app that we're deploying is from a localization standpoint where now we can deploy globally because we're solving the localization issues across the globe that large enterprises have. But we're allowing for the app now to be almost like that person's Persona into one organism of the enterprise by doing it that way. Now they only have access to the data points they're allowed to have access to. You don't have the problems with these large language models that literally are providing you access to HR files or financial documents or stuff they're not supposed to be in because they're not adhered to the actual security protocols that you have in place, why you need all these vaulted on solutions to it. So we at Maestro believe that the individuals who are empowering, yes, an enterprise is an enterprise, but individuals make that enterprise up. They're people too, right? So if you're able to provide an individual the ability to gain intelligence within a micro aspect and spark innovation, how you're really driving the momentum of a business to grow. And the way we've done this is through an application that they deploy on their machine. So the local CPU and GPU is actually leveraged first before it calls back to the backend and we call it a hive basically. And the backend is, is a simplistic server that, that doesn't need a lot of GPU because remember, Meister just serves. And instead of building models, we're serving the ability for the human to now create, create and save their thoughts as reasoning models to how they can now create collections to automate tasks down road or even interface with other AI tools that are in your enterprise. And by doing it that way, everyone is, we're leveraging every investment in place. But where Maestro first is, the most logical place to help consolidate a tech stack a little bit is the BI tools. Because if you look at the enterprise, BI tools at the end of the day are really not everyone's favorite thing to use. Let's be honest. They're more like in a technical extension to make you to let the business users understand how complex the technical folks have created all these great things for you to use. But the reality is it's not really serving the micro aspect of the organization. It's serving maybe the executive level or the leadership level, because they only can do six to seven different ones like Power. BI Tableau is what I'm really talking about right right now. And if you look at it from that angle, what are BI tools? They're basically graphics or data canvases to view your information through. Well, the Way that we design Maestro's app is to be a data canvas, which allows for the individuals to now explore through the data without ever leaving the canvas. Unlike other tools in the AI industry, where you have to talk to it and you pull it out and you start creating your logic, that's falling out of it, it's not falling forward. So allowing us to fall forward to exploring a data canvas is how you can now save the thoughts of how that individual solves that problem to the end result of the process. And this is how you truly can now intuitively and simplistically work with AI, because AI now can process your understanding of what you're trying to find, bring it back to you, like you explore through it, and provides you the ability to now save that. But if you think about it, there's avenues that you go on that where you might, might have an idea to go this way with the data and you start looking at it, but you decided to go back the other way, which we call almost like an Inception aspect. But that difference between those two is the complexity that you can't explain in an if and, or large language model approach, because they can't understand why you did that. That in Maestro's system, that's what we're really capturing, that uniqueness, because that uniqueness provides the humans understanding of the complexity of the data at that point or what they're trying to solve at that point. That's how you build a system that now can reason like a human, but still reward the human in the loop as a process of doing that.
- [33:18]
Yancy Sanford
It's providing you like an auditable approach to how you come up with a solution. And I always kind of talk to Chris and early days is that you've ever been in a meeting and you know, it's like, I don't have that information, I'll get back, right? Or like, if you're talking to like a process or something that you, you put together, you're able to like, branch off of a previous decision, to be nimble, to be able to pivot, and the ability to move real time in a process you already created, I think is extremely powerful. And then to add on to that, what we're looking at and moving towards is we're humans, we have what, five, six senses, depending on who you are or how you, what you believe. So your interactions with AI and intelligence, I see a future where it's not you sitting in front of a computer screen, it's more fluid, multimodal. Not so much like multimodal as in like a model would do an image versus audio but multimodal as in and human sensory and interactions.
- [34:27]
Chris Bush
What we're talking about is a world where data doesn't move. It's the human thoughts that move. And now you have the most secure system ever because at that point you don't have any of the data leakage data problems that current architecture today has. He's getting down to where the individual based on where the individual joins from a one organism standpoint of an enterprise or another business or whatever setting they're going into to it's their thoughts that come into that setting based on how they've applied those thoughts in other places. Similar to the data tagging of how those individuals interact with the data versus the data being the main point of what makes the logic. And that's where I think the paradigm comes into play where all these other systems out there today are really looking at the problem is like the data is a solution and what they're not understanding is the data is just, I mean it's just information. I mean it's, it's not, it's not the special sauce, it's the human. And if the human can, can freely flow through a situation that they're in to make better decisions faster. Now you've elevated human potential kind of.
- [35:38]
Yancy Sanford
Piggyback on what you're saying with the data. Like data is just an observer's measurement of something.
- [35:45]
Chris Bush
Right.
- [35:46]
Yancy Sanford
And like it just represents a thing. Whereas like to take apart kick back and not be a third party observer to data but to be able to intuitively. Yeah you get like a gut feeling or whatnot. And then a situation, maybe it's something with the data like points in like a certain direction that the human interpretation or gut feeling aligned with that can make you see that like yeah, that's the quickest route. But I'm going to be you know, jumping off a canyon if I go that way. So I know that's like not like that's just a rough thought example but yeah.
- [36:28]
Matthew DeMello
Self driving cars jumping out of canyons. This show is Chuck full of metaphors. Dan Fagella, our, our head of research loves to say I get paid by the anecdote in inter and I do think it's a very useful device to really communicate across different capabilities. But as I'm hearing it and everything that you guys are saying just in terms of how this works, like what you're saying just about data right here, it's not that you have all of this information based on outputs of that work what's really maybe more important is the logic by which the human being went from one step to the other to get that output. And here's your catch. What you guys see is the real ROI for businesses, that logic is best understood by the human. But with a system that best knows what to put in front of them next for that very next task and that's a much more agile system. A big problem I know Chris had mentioned before in one of the answers is that when you build out these systems to do reasoning for you, then they're trained on how your organization is as of we're recording this show on February the 2012th, 2025. And if you make a decision that changes your business goals on February 19th next week, well now you're stuck with that system that you paid all of this money for to do all of this reasoning for you that can't be changed because then you need to retrain it over again. Whereas a human being is very, very different. A human being is much, much more agile. So especially for industries where they see a lot of change and where you're not sure your business goals are good, going to last into the next five years or the next quarter or the next whenever, based on the size of your organization, it's going to be much, much more important to trust the intelligent operating systems we've trusted since the dawn of time of recorded history to handle these challenges. And that's basically human beings. Just want to make sure I've got that understood. I know we only have a limited amount of time before we head out of here today, but that sounds more or less where you guys really see, where you see the value especially for those organizations.
- [38:38]
Yancy Sanford
Well, yeah, so like just kind of as like if you think about it, life like the variables in life, they're real time and dynamic. So having a model that's stateful is not as adaptable as you in the process making it dynamic.
- [38:54]
Chris Bush
And I think what it comes down to is this is, is if you look at long term business success, right? And what, what organizations are really trying to achieve, they're in the process of building these systems and they think the way they're being told today this is how the system should work, right? And the problem is it's uncontrollable cost, right? If you built a system that allows for the enterprise to obtain almost like a proprietary brain of your employees micro aspect of how the data endpoint impacts the actual business growth, you're now rebuilding your knowledge repository of enterprise. And by doing it that way it's your dialect of what makes you successful in, in the, in society. Right. And why, why your business is thriving. That should belong to you at the end of the day. And that is really what we're getting down to. Because if you're capturing assist the system's capturing what you're what makes your secret sauce work, that system should be owned and belong to you not having to pay access for it down to street. And I think the problem is, is is we get lost in the coolness of technology versus versus actually understanding what that technology potentially could do to society. Right. And, and I think that, that if you have a system that's built to safeguard the individual in the digital world, where in essence your physical self when you bring a resume to someone is why you get hired every day. Right. And is why you actually get the job. You get it. But, but today, now that we have these, these AI tools that are, that are basically able to copy what you're doing and then you're what training your replacement, you no longer can, can protect your, your, your, your mental or your digital footprint in, in, in this world. And that's why we're seeing so many people laid off or that's why we're seeing people revolt against wanting to use a lot of these systems because they realize they are training their replacement and what's left for them them. So Maestro, at the end of the day is a system that's been built for the individual to now allow us to move in and out of our roles that we do and what we used to do when we used to go to each company and gives us the ability to now think of the world of more of like intelligence as a subscription and selling our intelligence to be able to work for multiple organizations. Because let's be honest, it's only a matter of time that these organizations are going to start laying off employees more and more and more because they're automating everything. And at that point they've basically taken the intellect that made you valuable. Now we're how do you, how do you make money in the world? And that's why we're talking about jobless workforces. Right? With Maestro, we don't believe there's going to be a jobless workforce. We believe that the next evolution of collective human intelligence is going to be driven by, by us selling our intellect to multiple organizations. And instead the organizations also get what they want where they get lower rap rates and become more competitive. But the human is always still rewarded in the loop, which truly makes the greater good for Collective human intelligence versus the few. Just owning it.
- [42:14]
Matthew DeMello
All right, so two things there. It's less about, about the tribal knowledge, which, which I thought you had done. I was going to ask in the middle there, hey, what's the difference between tribal knowledge? But I think you had really differentiated through your answers. The big difference there. But if we can put it in a nutshell based on your last answer, it's really more individuated knowledge of hey, here are your skills, here are your skills that you're going to take from company to company to company. And here is the presentation of the data you need to enhance those skills and to do the job you were born to do, all the better, because we know the trust is in your hand. Really fascinating stuff here, Chris, and I'm really glad you chimed in at the end of it. The other half of that is you're talking about proprietary data. A big motivator, at least for your OpenAI's. All of your big tech names that are in these fields is they do want to drink the data about our organizations, about our lives, about every facet of what we do in for the themselves. But I think that speaks to another dynamic of what we're gonna see play out in the market is right along the lines of what you said in terms of employees. Why do I wanna participate in a tribal knowledge system where I'm training some system to take my job away from me? It's gonna be the same thing of companies starting to ask themselves, do I Wanna really give OpenAI all my data and how I think about my business goals. That's a little personal. Do they really need that aside from building me a better tool? And I think those are two questions that we're really going to see play out in the market over the, over the next few years and really come to the fore. So really appreciate these conversations. Chris and Yancy, thank you so much for being with us this week and really, really having some high level conversations about this.
- [43:59]
Chris Bush
Thank you.
- [44:07]
Matthew DeMello
Lots to pull from from today's episode. But before we wrap things up, I want to put a bigger highlight around three key points I think Chris and Yancy brought in to today's show. First, enterprises are racing to build AI reasoning models. But the real value lies in augmenting human intelligence, not replacing it. AI should enhance decision making, not take the wheel away from the people who understand their business the best. Secondly, over reliance on AI can lead companies into an uncanny valley of decision making where employees disengage and business agility suffers. Keeping humans in the loop ensures adaptability, trust and long term innovation. And finally, the future of AI isn't just about automation, it's about capturing human expertise and reasoning at scale. Organizations that prioritize AI as a tool for enabling human potential will see greater ROI and maintain a competitive advantage. If you enjoyed or benefited from the insights of today's episode, consider leaving us a review on Apple Podcasts. Let us know what you learned, found helpful or just liked most about the show. Also, don't forget to follow us on X formerly known as Twitter Merge and that's spelled E M E R J as well as our LinkedIn page. I'm your host Matthew DeMello, editorial director here at Emerge AI Research, on behalf of Daniel Fagella, our CEO and head of Research as well as the rest of the team here at Emerge. Thanks so much for joining us today and we'll catch you next time on the AI in Business podcast.