Inspired Execution

Chet Kapoor: Stephen, welcome to the podcast. Thank you very much for having me, Chet. So you've had a phenomenal career spanning government, manufacturing, education. Your focus is on cognitive computing, helping computers think and learn in ways similar to humans. How did you get started in this field? 

Stephen DeAngelis: Well, I got started in this field by solving problems in analysis, in solving problems and optimization as a governmental contractor. So in the early 2000s, right after September 11th, we started to work with the U.S. government in performing, connecting the dot type problem solving cases back then. And we needed to have a way of analyzing tremendous amounts of data very quickly where you couldn't, you didn't have the ability to have enough human-based analysts process the data fast enough within a decision cycle. So we needed to find some help. So we entered into the area of artificial intelligence, not to set out to build an AI firm, but to build a company that was solving problems in counterterrorism. And we needed AI as an active tool to help with that. 

Chet Kapoor: That's awesome. That's awesome. And how is that going so far? Was that a, is it like, has it gotten a lot bigger? How is that coming along? 

Stephen DeAngelis: So the journey for our company, I think parallels the journey of AI, right? Where you have seen explosive growth in AI in the last several years. And that growth has primarily come from the use of generative AI tools to popularize and democratize people's understanding of artificial intelligence. AI has, sort of two main branches, as you noted, Chad. It has human-like reasoning or symbolic AI, and it has generative AI. And symbolic AI is very hard. It's, requires the use of mathematical logic to perform forward chaining and backward chaining of mathematical logic to understand data much like a human would. Generative AI evolves out of search capabilities and is a lot more scalable, but sometimes less robust today than symbolic AI is. So, one requires a little bit more heavy lifting and the other is a scale play, and hopefully those two will converge very shortly. 

Chet Kapoor: Yeah, it seems like people are making progress, right? In making that happen. So you've been recognized by Esquire magazine and Forbes for your innovations. What was, if you look back and say, what was a pivotal moment that actually shaped your career? Or, it's not a moment, right? But, what was the one thing that really changed your path, your trajectory? 

Stephen DeAngelis: When in the aftermath of September 11th, the government was looking for distinct applications that could help solve problems. That market driver back then of there's a large problem that we need to rally around and solve, created a lot of energy or forward inertia in the ability to, forward motion rather, in the ability to drive investment into tools that could help perform analysis to keep the United States safe. Similarly, in the late, similarly, when data, when the storage and access of data through cloud computing environments became ubiquitous and you, in the cost of storing huge amounts of data, essentially became free or very low cost. That also drove massive investments in tools to be able to help utilize that data. I found that in the early 2000s, the critical stimulus was, how do we stop bad guys from doing bad things, right? Yeah. In the 2010s, it was about, we have a ton of data. How do we have the analytic capabilities catch up with the data obtaining that we have been very effective in since 2000? Those two events created conditions into which people can drive investment against these problems. And that allowed AI to flourish at a time where software engineering was also becoming much more scalable and mature with, Python and other coding techniques that were, becoming easier for a broader base of coders to be able to access. 

Chet Kapoor: Yeah. No, that's awesome. That's awesome. Let's talk about your company. Enterra Solutions helps enterprises with value chain optimizations with your platform, right? And you have awesome customers like Nestle, Kellogg's PwC, Salesforce, so many more. What specific problem are your customers facing? Or what are the key problems that your customer faces? 

Stephen DeAngelis: Our customers are consumer products firms, right? The PwC's, Accenture's are partners of ours, right? And what happens is we have, we're out solving what we call end-to-end value chain optimization and decision-making through a system of intelligence. So, when we came from the commercial government, rather, to the commercial sector, Chet, we identified a huge marketplace for a thing called a system of intelligence. Think of it as the counterparty to the transactional systems of record. The SAP's, the Oracle's, the IBM's, the Salesforce's, right? And we designed a business, Fish for Purpose, a technology platform and a set of business applications that would fill that market gap for a system of intelligence. 

Chet Kapoor: That's great, that's awesome. And what technology have you created to make that happen? And by the way, I love the system of intelligence because the world I come from, there's always a system of record and a system of engagement, right? This is how developers build apps and things like that. So, we always call it two-tier systems, right? And I clear that there's a system of intelligence that has emerged in a massive way. And in that, you can put LLMs and a bunch of other things like that as well, RAG-based systems in addition to everything else. But what technology are you using to solve those problems for customers like Kellogg's and Nestle? 

Stephen DeAngelis: Yeah, so let me just go back to the answer to the previous question in a more detailed way because it'll help explain it to your audiences. You asked me about what problems we're helping them solve. If you think about a system of intelligence, it spans marketing, where we look at how do we perform high-dimensional attribution of a consumer preference to a product. Think of it as consumer insights. The sales function, it's holistic revenue growth management. How do you drive trade promotion, pricing, assortment, media mix, and then supply chain? How do you bring it? If you look at solving problems along that continuum, and then we have a cool, and those are prescriptive analytics. And then sitting on top of that is an anticipatory. Think about like sonar and radar for a company in a solution we call business wargaming, which allows companies to use game theory to simulate the nature of competition in the marketplace, understand the conditions under which they take share from competitors, or competitors take share from them, and then plot the optimal play you can run to beat the competition in that marketplace. So, those applications were important because in order to sell artificial intelligence, you need to contextualize the AI to a set of very rich use cases that drive profit, right? These are very compelling use cases that, now, the technology that enables that is a platform that has three pieces, a human-like reasoning, artificial intelligence, and generative AI combined with glass box explanatory mathematics. You're probably familiar with machine learning algorithms, they generate patterns, right? But you can't introspect the pattern. Here, it's a glass box engine that allows you to generate a function or a combination of variables that explain what you observe in the underlying data set. It's like an x-ray for the data. It says, oh, I looked at this large data set, here are the variables, the 5% subset of the variables that drive 95% of the variation in the data. And then the third piece is a non-linear optimization capability to optimize truly big pieces of the value chain. That's a platform, human-like and generative AI combined with glass box mathematics, combined with non-linear optimization can constitute an autonomous decision science platform that then enables the business applications I described a minute ago. 

Chet Kapoor: That's awesome, that is really great. So, you've been collaborating with educational institutions as a researcher and a lecturer, right? You've done this for a while. Yes. I loved your quote, right? AI is a net job creator. Two questions, what types of roles and skills are emerging in this new economy, this new AI economy that we are going to see over the next few years? 

Stephen DeAngelis: Well, I think, so Chet, I think there's a pyramid of skills, right? A mass job creation that's gonna come from AI-enabled roles. There will be foundational roles for people who are involved in building infrastructure. Anywhere from, you know, the, if you look at Microsoft and OpenAI's one investment of $100 billion into one data center in Texas, right? Think about the scale of investment. It's almost like Intel investing in a fabrication facility. Huge amount of money going in to construction. Energy investment to fuel the data center. Having to build power plants to have enough electricity to fuel the data center. NVIDIA chip manufacturers to make the chips to power the servers that are in there. People who code the software to run the data center. People who build the large language models, right? To operate it. So, I think you'll see a ecosystem of jobs, right? Not just, AI jobs, specifically in, prompt generation or things like that. But you'll see a whole ecosystem of infrastructure jobs. And then within the AI sector, you'll see a pyramid. And you'll have, think about the large base of the pyramid being people who can earn a middle-class living, performing vocational or light, heavy, light educational driven skill sets who are practitioners and in maintaining and operating the large language models. Then you'll have a class above that who are people who have, college degrees or master's degrees or have been in the field for a while who have enhanced skills. And then the top of the pyramid, you'll have, PhDs or others who are the architects of how the systems will work. And then on top of that, you have, geniuses who invent this stuff, right? In the first place. But that pyramid is, I think, can be the new middle-class job creator for jobs that have been often offshored over the last 30 years in the manufacturing sector. So I think it's going to be a very compelling ecosystem of roles that we're going to be unleashing with AI. 

Chet Kapoor: Let me ask you a question. One of the things that I talk about when I have a chance is, we've been living with AI agents for longer than most people think, right? I mean, there's an AI agent called Google Maps on this. By the way, it is powered by predictive models, right? Based on how many people are going in certain places and a bunch of those things. And we've been, you know, we've been dealing, I mean, think about it. My life would not be that great without my Maps agent, right? And so there is this concept that I'm more productive with an agent, with an AI agent like Google Maps. That is in your first year as a consumer or the second tier? 

Stephen DeAngelis: The people who engineer it are across all tiers, right? But the users are in the bottom tier. Users will be on the bottom tier, right? Correct. Now think about other things like autopilot on an airplane. Yeah. You trust that, right? People say, I remember, I wasn't around, but when autopilot was first introduced, you know, people might be skeptical about autopilot, right? But the notion is that people have come to, or self-driving cars, right? So the concept of autonomous systems, which is what our company really, if you ask me what is the core DNA, it's constructing autonomous systems for industrial scale application, right? Yeah, for sure. Those systems have been around, as you mentioned, for decades now in various shapes and forms and maturity levels, right? Correct. At the end of the day, there's going to be a boom in the use of agents. Now there's going to be a whole set of things around agents, Chet, as we start, in terms of ethical operations and governance and security and all that stuff. There's a lot there. 

Chet Kapoor: And I think we'll figure that out. I continued as I've interacted with customers and even gone to the World Economic Forum, what I've seen is, this is a societal problem, right? We have to figure this out at the team level, at the company level, at a national level, and then at a global level, right? Because I think it cuts across all of those, right? How do you do ethical AI? How do you do responsible AI? And all those things. I think you would agree with that comment.

Stephen DeAngelis: I totally agree. And, if you look at, just the concept of information, disinformation, right? Correct. Large language models, right? Are going to evolve. If they evolve ubiquitously, they're going to have to separate disinformation 

For real information. Correct. You can't have garbage in and garbage out on steroids. So the question is, how does that happen? And I think that's going to be the subject of probably hundreds of billions of dollars or more investment by large tech firms to figure out the best way 

Chet Kapoor: Correct. And I personally think that one of the, I actually am very hopeful. And, every technology in the world, every technology that's ever happened in the history of mankind has always had negative, there's a negative aspect to it, right? It's not just the positive part, whether it was Steam, the internet, mobile, right? Everything has had a negative aspect to it, right? But I do believe that this is the first time in the last four waves that I've been involved in, where the government is getting involved sooner rather than later in trying to figure out. Now, there's good news to that, but there's bad news if you don't have the right people in government doing it, right? Because you need somebody who understands the technology as well as understands what the next generation that grows up is going to grow up with, not the generation that is in their 60s and 70s and 80s, right? It is for the future generations, as I call them, born on AI, right? As I used to say, born on web. 

Stephen DeAngelis: I agree with you. And I think that understanding who are the right people to play these roles, much like in your company, you try to hire the best person for the role that you find. The government has a harder job because they've got to try to find the best people without being able to give them the same compensation that they would have gotten in the commercial sector. So you find people, you want to find people who are super competent, but also mission-driven to help with this problem set. But I'm also increasingly hopeful. I think one of the reasons you see them more actively involved now is that people are acutely aware of the information, disinformation, things that are going on in social media and elsewhere. 

Chet Kapoor: For sure. So question for you on future trends, what advice would you give to students who are entering the workforce? 

Stephen DeAngelis: I fully believe that if people can get as many hard skills as they can obtain, like it really matters right now? Yeah, I agree. So engineering skills or any discipline, it doesn't have to be computer science, right? So if I'm going to choice, am I going to hire someone who has a degree in applied mathematics versus someone who has a generalist degree, I would likely hire the applied mathematician. Or the, because I feel that the technical skills or the engineering skills give you problem-solving capabilities. 

Chet Kapoor: Correct. 

Stephen DeAngelis: Allow you to address a problem, design an experiment, do things according to a scientific method that, if you're in, and I'm commenting on my end of the world, right? So the artificial intelligence, right? For sure. The obtaining of more hard skills, I think is going to be preferred than softer skills. Because companies will train you in the softer skill, but you need to come to the table with the right tools in the toolbox. 

Chet Kapoor: Yeah, no, I 100% agree. Can I see if you actually agree with this small modification? Sure. Figure out a way to get educated on structured thinking. Yeah. And obviously that, you know, hard skills is structured thinking plus plus. But, my take is, go to statistics if you don't want to go into, but try things where you learn how to approach a problem in a very structured way. Large. Versus, because you need the arts as well, just to be clear, like you said. Because it's a combination, I mean, great software, as you well know, Stephen, is created by people who, you know, with liberal arts and computer science, right? It's not just done by computer science problems, right? So would you agree with that? Structured thinking is a critical, should be a critical part of your education? 

Stephen DeAngelis: Well, that's what I meant by hard skills, right? It really is if you're logical thinking, problem solving skill sets, right? Allow you to dimensionalize a problem and then create a structured way of solving it. 

Chet Kapoor: All right. One bold prediction for the next two years of, in two years for Gen AI. 

Stephen DeAngelis: I think Gen AI will start to become geometrically embedded into the tools, your phone, all the tools of today. It will start to get further ubiquitously embedded and hopefully it'll start to have a trust factor. You'll be able to start to build trusted LLMs versus LLMs that are still casting wide nets and bringing in all the information. 

Chet Kapoor: That's awesome. I would agree with that 100% of everything you said. All right, we're at the rapid fire stage. Okay. So first question, what book has had the biggest impact on you recently? 

Stephen DeAngelis: Recently, you know, recently and permanently, there's a great book called Gertl Escherbach, right? By David Hofstadter, Douglas Hofstadter rather. It is, it brings together logic, design and creativity music in to what they call an eternal braid. So it's a very cool book. I recommend it. Douglas Hofstadter, Gertl Escherbach. 

Chet Kapoor: I will have to get it. I have never heard of it and it'll be on my Kindle this afternoon. So photography is one of your hobbies. What's your favorite subject to photograph? Nature, 

Stephen DeAngelis: as you can see where I'm sitting, right? I know. Nature is my- 

Chet Kapoor: On your spread. Which industry will see the most transformation due to AI in the next five years? 

Stephen DeAngelis: I don't think it's gonna be, I think it's gonna permeate almost every industry, right? But I think the tech industry, obviously. But aside from tech, I think it'll be across any of the manufacturing sectors. You already have a lot of it in motion pictures and entertainment today. So I think you will see it transform almost every industry. But the face of technology, I think, is going to dramatically change. 

Chet Kapoor: What's one skill you think that AI will never truly master? 

Stephen DeAngelis: I don't think it will ever master the creative side of the human's aspirations, hopes, dreams, and the embodiment of that in the creative spirit of a human. 

Chet Kapoor: Yeah, I would agree with that 100%. What three words describe the best leaders? 

Stephen DeAngelis: I would say visionary, right? I would say trustworthy, right? And I would say the ability to manage teams. So vision, trustworthiness, and management capability. 

Chet Kapoor: That's awesome. Stephen, this has been phenomenal. Thank you very, very much for your time. I think our listeners will absolutely, our audience will absolutely love this. We really appreciate you coming on. 

Stephen DeAngelis: Thank you for having me. I really appreciate it as well.