0:00

Andy Sack: Welcome to AI first with Adam and Andy, the show that brings you straight to the front lines of AI innovation and business. I'm Andy Sack alongside with my co host Adam Brotman. We have conversations with business leaders who are transforming their business with AI. No fluff, straight talk, real use cases and insights for you. Welcome, Nick Soman. Adam, greetings. Nick Soman: Thank you. Nice to be here with you guys. Andy Sack: Hey Nick, super happy to have you here. I've known Nick for a long time. Uh, he's also been a customer of Forum three. Uh, Nick, why don't we start out with you just explaining to people, give people a brief introduction to your business. Decent. Nick Soman: Sure. So my business, Decent, builds health plans for small employers around this new growing care model called direct primary care. The cost of employer sponsored health coverage has quadrupled since the year 2000. And everybody who's in the business knows that people aren't happy with what they're getting from health insurance companies. Now the big companies in America have all figured this out years ago and they'll do something called self funding where they are able to take more risk themselves and provide better coverage at a lower cost. What decent does is we bring those tools down to the little guy so we can help companies as small as 5 employees self fund. And we do it around this model called direct primary care, which to you as a patient or a member means unlimited primary care with your personal doctor. We're giving you what feels like concierge care at a lower price point and we're growing really fast and right now everybody's happy. Andy Sack: It's been a journey for you, uh, what's the size of the, of the company and, and your role there. Nick Soman: I'm the CEO of Decent and I'm the co founder with my other co founder Richard. There are about 30 of us now full time and we're evenly distributed between India, the United States and the Philippines. Andy Sack: We're here today to talk about your journey as both a executive, uh, CEO and your journey as a company in really just using uh, AI and when you use it, what's worked about it, et cetera. So give us a little bit of a high level of just your own personal usage of AI and sort of when you decide when you had an aha moment and how you decided to start, uh, to use it in business. And Adam, I want to invite you to try and, you know, we're working to try and make this a, uh, conversation, so feel free to dive uh, in with both comments and questions. Nick Soman: I use AI constantly. I actually spent yesterday Morning. Uh, running my own sort of Wet test with AI where I use ChatGPT, I use Claude, I use Perplexity, and I used uh, Gemini. And I ran them through five different tests that are all relevant for decent and then just took notes for myself and my team of what I observed. So I'm at a point now where there's enough variation in the tools that I'm starting to use, uh, more than one tool for most of the AI tasks that I do. And how I use AI is a really good microcosm for how my whole team is starting to use AI. I use AI to do more of the work so I can focus my effort on where I am highest leverage. And at decent, we work in this really archaic janky space with these gigantic incumbents that are, as an industry the single largest employer in America, which increases the cost of what they sell drastically every year. And what we realized is, hey, we might be able to build the first billion dollar company with, you know, 20 employees or less in the United States if we use AI really thoughtfully and then we bring in some people from, uh, other countries to do full time work as well. And so what I'm doing and what my team's doing is figuring out if you think about it as a workflow, can technology do it that now includes AI? If so, we're absolutely going to push it there. If not, can somebody who lives outside the United States do it? Could that person be part time? If not, could they be full time? If not, could they be part time in the United States? If not, I guess they could be full time in the United States. But we really run everything through that, that thought pattern. And it is incredible how fast AI has been evolving to take more and more of the work. Adam Brotman: Nick, what's an example? I mean, you don't have to be say anything confidential, but like, what's a high level example of the kind of workflow that you and your team used to have to do yourself manually that now you can automate or speed up or improve through AI? Nick Soman: I, uh, don't want to plug Forum three too much, but I will a little bit. So you all got us really thinking about how we could create uh, our own AI tools that have context about all the things we do at Decent. The thing that's pretty mind blowing is, you know, Decent's had a journey like Andy said, and he's been 00:05:00 Nick Soman: part of that journey. We've had times when we were more than or roughly 100 employees in the United States and we had a big old Talented marketing department as an example. And a good example of a workflow is now with these multiple AI tools. I have fractional, uh, CMO Rachel Kim. She's absolutely fantastic. I used to work with her at Gusto. She and I work closely with our product leader who has innate interest in marketing, and we work with our broker, sales leader, Chantal. But I would say at this point, AI is doing the majority of the work to sort of understand the context of what we do. And we'll come to it with really increasingly nascent concepts. We'll say, hey, you know, we're thinking it might make sense for us to consider offering a health share alongside our health plans so that we can serve individuals and certain small businesses. How would that work from a marketing perspective? And the difference is really, it produces a document that's in the format that we like, that is contextual and aware of the different decisions that have come before. And we're able to take it as a first review. And as anybody who's ever written anything knows that first review and that edit, it takes a significantly less activation energy than sort of working from a blank page. And I think that's, that's probably at the high end. A lot of what we do is just, you want new copy on a website, you're going to go to AI, uh, you want to figure out how to put together a pool of answers that can have enough contextual awareness that when members ask how our plan works, they're giving accurate answers while also recognizing that every member is a little different. This is something we do with AI. I think, um, what is mind blowing to me is this. If you had told me seven years ago that we would be, I think, at the forefront of how people are using AI in the health plan space, I would have said, no, no, no, no, no. That's not the kind of company I want to build. I want to build a company that's really about putting people back into the feeling of having a health plan and people back into the feeling of having healthcare, not computers. And what was wrong in my thinking was it turns out that AI at its best, I think is an Iron man suit. And we use that analogy constantly, internally at decent. We're able to do more with the talented people we have, uh, because we're using these technologies to extend our reach. Andy Sack: Nick, I'm curious, is everyone in the company using, uh, AI? And do you have a corporate tool that you subscribe to? Nick Soman: We have an enterprise, uh, instantiation of ChatGPT. And then we also will fund people who My, my marketing leader likes to use Claude. She thinks it's better for certain aspects of writing. And so, you know, it's funny, you become cash constrained as a startup and you worry about these things. And I appreciate, and I love that my team knows that I'm likely to say, oh, could we do this for less? Could we do this for cheaper? AI is one of these tools that I honestly think is drastically underpriced right now for the value that it's creating. And so I've gone to my team and said, look, if it's going to cost you 80 bucks a month to use all four of these key tools, get them all, we're not going to go buy them for everybody because you need to earn the right to use them by showing that you actually will. But it's a no brainer for us because if you think about the cost, financial and emotional of hiring someone junior who doesn't really know how to do the job and needs to be managed, I mean if you could take an $80,000 slug and that's on the low end and cover infinite AI tools for everybody at the team who wants them, that's a no brainer. We've recently started to introduce AI for our international teams as well. And um, it's, it's amazing how quickly they're getting up to speed. Adam Brotman: Do you find yourself running like an AI council with like AI office hours? How do you, you're, you know, you're not a huge team but you have enough of you that are, are you sharing best practices in some ways and kind of tips. Nick Soman: And we are, we are, we have a dedicated Slack channel, uh, that's around how we use AI thoughtfully. The other thing, and this is going to sound a little callous but I don't care, I can be a little callous. I've gone to people and I've said, look, AI is not going to take your job unless you don't use it. Somebody who is using AI, and I didn't quote this but somebody else said it and they said it well, is going to take your job. If you're not serious. This is a job, right? We're having fun here together, we're growing, it's really exciting. I expect you to do your best work and when new tools become available and we're happy to pay for them, get in there, give it a try. And thankfully my team has been really willing to do that. But I have set the expectation that if you're doing things manually and someone else points out to you, hey, look, there's a way you could significantly reduce your workflow so you can focus more of your energy on the tasks where decent really needs you to be yourself. You know, most of the time people better have a darn good reason if they're not willing to do it. We don't comport with people that are just a little uncomfortable. Adam Brotman: Yeah, that's, that's cool that you're talking. I liked your Iron man suit, but did you say you think of it as like wearing an Iron man suit? Kind of. Is that what you said? Nick Soman: It's the Iron man suit for everybody. I mean it's, it's Jarvis too, but, but this is the thing I needed to sell to my team because ultimately every, uh, exercise in change management is a sales job. Right? And I had to go to them and say, look, I'll think about Julia on my team who's absolutely incredible. So, hey Julia, if you're listening to this, 00:10:00 Nick Soman: this is the woman that runs our nurse care management team. I am under strict orders from my wife Anna to give her whatever she wants because she takes care of our family and she's absolutely fantastic. And she gets to be fantastic by being a human. And understandably, anybody that's worked in a large, ah, organization which she has, where the humanity was sort of lost, is going to be reticent to embrace new technology. But I was able to kind of go to her and say, let's try out these tools. And she can now see how, how much more she can do without working late nights, uh, by virtue of being able to set up a system that can at least take a first shot at answering members questions so that, you know, you only need the human to kind of approve the answer before it goes out as an example, which I think, by the way, is going to be one of the major heuristics in healthcare in the near term, uh, there's a lot that, um, she can do now that, that she was feeling very overwhelmed by before we started to step into this world. Adam Brotman: Yeah, we, by the way on that. The reason I brought the Ironman suit is that I love that analogy of yours. And I, my reason I brought it back up is you said it's sort of like, I, uh, like how you put it. You know, people should be using AI, but they should be kind of using it all the time. Like, and it's something that we, we practice, we, we preach that. I should say we, uh, we practice it too. We probably don't practice it enough to be honest, but we, we, we, you know, Andy and I and our little team, we use AI dozens of times a day. So we're using it a lot. But, but to be honest, it'd be like the iron man suit. It'd be like Tony Stark going out and fighting crime without his iron man suit would kind of be foolish. Like, he might be able to do it, but like, but why would he do it? Yeah, it is amazing because like, I don't know about you, but I have found myself Andy and Andy's a big fan of. I'd love to get Andy chime in. Like, like I'm. Now I go into meetings, conversations, hoping that they're being recorded. Which is weird, right? Nick Soman: Because we use fathom for everything for that reason too. Yep. Adam Brotman: Yeah, because it's like, it's like, okay, if I want to wear my iron man suit, I just want everything to be kind of running through the AI and like, and being just step changed, improved. And if you're not recording it or you're, you know, you, you sort of forget sometimes, like, oh, I, I, you don't have to record it. You could just relay a conversation, take your notes and relay them to the AI. But it's, it's going to be. Andy Sack: Adam, explain, explain why, uh, why you want to record everything. Because I'm a huge, Obviously, you know, I am. But just to connect that for people who don't know. Why is that important? Adam Brotman: Well, I mean, we, it's fascinating that, you know, you're, um, if you take the transcript, okay, so if you're recording everything, there's going to be a transcript, which means you have, you've essentially created a data set of a conversation and then you can take that data set, you feed it to the AI and you can do anything you want with it. You can say, make it better. Uh, give me a summary. Summary is like the least impactful thing you're going to do. You're actually going to say, analyze this conversation. Tell me from this conversation, like, what insights were m. You know, what questions did we not ask? What are the next steps that we should be focused on and prioritizing? Like, it's incredible. If you do that for every meeting, all day, every day, you're going to see a 50% increase in quality and productivity across everything you do. And like, you know, that's incredible, right? It's an incredible gift. It seems, it seems like it's not possible, but it is. So that's why it's turning conversations into data sets and then letting the AI be that iron man suit for you. Nick Soman: Uh, it's also context that makes you smarter every single time. My favorite test that I did with these four different AI tools, and the responses were funny. You know, we're doing founder led sales, which, by the way, Andy, I love it. Oh my God, am I having fun over here doing this stuff. And there's still room for Chantal, who's incredible and she's doing all the relationship management, but nobody can get up there and talk about this thing like I can because I've been doing it for seven years. And so I have been leaning into what that looks like going into the new year. You know, Q4 is a heavy sales season. We've gotten some ideas, including from AI, on how to crack seasonality in our business. But I wanted to look at how do we scale kind of our product led, our, uh, founder led sales. And I have a couple friends who do this for a living. One of them is Pete Kazanji, who's a good friend of mine from Stanford who wrote the book called Founding Sales. Another one is a woman called Jan Abel. She's probably the other person on Twitter that's best at early stage sales. And I emailed both of them to say, can you take a little time with me? I got bored of waiting for them to say yes, which they ultimately did. And so I went to AI and I said, hey, I just asked this question to Pete Kazanschi and Jen Abel and tell it anything. I actually said, imagine if Pete Kazanshi and Jen Abel had a baby. And it was funny because ChatGP immediately picked up what was desired there. Claude said, I cannot speculate on the personal interests of people. His Claude is too high up the ethics ass, uh, and perplexed that he couldn't do it and connected it with the wrong gen Abel, who I guess is a Canadian diver. But as soon as I corrected it and said, you tell me what they're going to say, it wrote this incredible summary. 00:15:00 Nick Soman: Then I did that with three other AI tools. Then I asked the AI where it disagreed with itself. Then I synthesized it and then I sent the answers to Jen and Pete and basically said, this is what it thinks you're going to say. And they both said, yeah, that's pretty much right. So now we can use this time that we've scheduled to dig into the next layer of questions, which, that's mind blowing to me. I would not have been able to do that in my imagination, uh, three years ago. So I'm right there with you. That picking up that context on us and the world around us is Everything. Andy Sack: That little vignette is a great example of the, of the Iron man suit. Right? Nick Soman: Like, oh, yeah. I mean, why should I wait for somebody to tell me what they think when I can get 95 fidelity by asking a computer and then they can correct it. Adam Brotman: Yeah. Andy, doesn't it remind you of what Joe said on. We had another, uh. Andy Sack: He knows Joe. Nick Soman: Is this my Joe? Yeah, I like Joe. Adam Brotman: Yeah. Nick Soman: Okay. Adam Brotman: And, uh, we had another CEO of a startup company that was, you know, like you. Nick was really, you know, using AI all the time. And he gave an example of talking, doing a fundraising round, I think, and he was saying, uh, well, I'm going to go off and I'm going to basically do kind of what you just said with all the different, uh, venture capitalists that I want to go talk to, and gave him the context of prior conversations combined with the research AI could do and effectively was like, tell me what they're going to say when I go. Nick Soman: Uh. Andy Sack: Right. Adam Brotman: Which is an amazing way to prep for a meeting. Right. Or, like, prep for an interview or whatever. Like, it's, um, it's. It's great stuff. Nick Soman: I actually want to give one more, one more quick example. And this is a short one. Yeah, this is something I did yesterday. Andy Sack: Examples are. Examples are great. Nick Soman: Yeah, I, I had a meeting with a friend of mine and we're looking at doing something pretty interesting together that probably connects pretty deeply to Decent's next big play. Not that we have. We have so much we can do on the current big play, but I'm always looking at the future as well, because the ambition of this company is affordable healthcare for all. And I mean that. And so I was talking to a friend of mine who can help us get there. And I sensed that the conversation was going well, but there was a little bit of, uh, an undertone that I couldn't identify. So I fed the transcript into ChatGPT and I said, what's going on here? What's the relationship between these two people? I had literally just been in the meeting and it was able to summarize with specific examples, areas where my friend was coming off as, you know, a little conservative and wanting to make sure that I was going into this eyes wide open. And I even argued with the AI and it said, well, if you look over here, you know, this is how he reacted when you said this thing, you know, and then he said this thing. And I had no, no notes. I had to go back and go, okay, you're absolutely right. There's a, uh, a conservativism here. And there's sort of a wariness, not in a negative way, but that became part of my follow up to the guy is to kind of go, hey, I can, you know, I, I can read people pretty well in general, but having that as a tool at my disposal, it, it put a name to something that also probably would have been a little awkward for me personally to name, you know, without help. And I was able to use that to kind of lubricate the, the relationship going forward. So the closer this thing gets to understanding how people think, how they work, how they, um, bluntly, the less people we're going to need to, to build big companies. Andy Sack: Nick, you mentioned earlier in this conversation that you think you said something along the lines of that, that AI tools are underpriced. I think that was what you said. Talk about the value versus the cost in your mind and then also talk about where you would like to, where you'd like to see decent go in your AI journey. Nick Soman: So the reason that it's hard to put a value on AI in the early stage startup world, I think is related to the fact that we've all gotten it drummed into our heads that the way to build a team early on is to specialize, specialize, specialize. You don't want people hanging out that are just pretty good at a lot of things for the most part. You might want a couple of them early on, but that's really as fast as possible. You want to hire the world's best designer that'll work with you as a founder, you want to hire the world's best technologist, you want a world class CEO. And AI sort of turns that paradigm on its head because it's almost like if you could get a fractional genius for, uh, every part of your company at the same time, how much would you pay for that? And the answer to that isn't clear because I can't just say, well, cost me this to hire a junior person or it cost me this to hire a senior person for one function, right? The idea that you could have somebody that consistently adds 25 to 50% productivity and thoughtfulness to your decisions, nobody knows how to price that, right? And I think that's only going to get better over time. What I've told my team, because that's probably the best arbiter of truth. I've told them, hey, we're not there yet. I want to be Netflix. I want to build a white Knight version of UnitedHealthcare. But culturally, I want to be Netflix, I want to know we've got the absolute best people on this team in the United States. I want getting a full time job at Decent to feel like getting a tenure track job at Stanford. Like you could leave but you probably won't. And I think that if you can get the right people and they have that growth mindset that is really predictive of whether they'll use AI, uh, the right way, 00:20:00 Nick Soman: uh, I think this is the direction things go. I actually, I'm practicing what I preached. I just reached out to a friend and he had said, uh, the best people leader I know just became available for fractional work. Now what's interesting is I thought, okay, I don't really need that right now, but I'm anticipating we're going to be growing in the not too distant future just because the business is growing. I wouldn't have said this two years ago, but I feel like a great fractional thinker supported by AI and some of the people we're working with in the Philippines as an example, could probably do a better job than somebody I could get and pay a good rate for full time. It really changes the game. It's how can you have access to the best original thoughts and then use the AI as a scaffolding and use maybe some more junior talent to sort of do the day to day work. And that's an incredible inversion, especially for someone that loves working with leaders. Andy Sack: Talk about, ah, uh, that's the second part of my question which was where you want to take, where you see using AI going forward, like where you, where is decent in its AI journey and where do you see going? You, you talked about, just prior to recording, you were talking about your mouth being bigger than your stomach or. Nick Soman: Yeah, my eyes are being bigger than my stomach. So this is the nature of being a founder, right? I look at this and I'm like, look, let me break it down for you. So a, uh, health plan is a math problem. It's really, first it's an adoption problem and then it's a math problem. The adoption problem is can you get people to buy your thing? And then the math problem is can you, can you make enough money that you can serve them without, without going broke? And, and it is dead obvious when you break down health plans, which by the way are Warren Buffett's favorite investment, they are some of the biggest companies in the world. They are gigantic. It's not complicated when you look at it at a root cause level. Now category AI help with adoption, which is my side of the House, you're goddamn right it can. Can AI look at things like find all of the information that's available for every employee of this company on the Internet and use our existing actuarial tools to predict what we're going to be able to give them a plan for. Since we use all of this standard human information today, do that assessment anyway. How close can we get if we've got the world's best AI tools? So I can go to somebody and say, maybe, I don't know that I can save you 40% on your existing plan, but I'm confident enough based on running a bunch of people through this and looking at the results that I'll give you 100 bucks straight up. If I can't, things become possible that weren't possible before. Because you just think about it like, like Adam said is it's a new creative way to be able to parse the entirety of the Internet all at the same time and um, add that to the context of what it knows about you. So that's on the adoption side. On the math side, you know, I could hire a huge data team and it probably wouldn't be able to do what ChatGPT can do today. Looking at our systems and figuring out for this group, you know, let's, let's look in the moment then let's look in the future in the moment. Which of our groups are costing us money? Why are they costing us money? Are we missing any near term opportunities to be able to better serve those groups for less money so we can give them a great renewal next year? And then you can easily start to see, okay, it turns out groups like this tend to cost us money, whereas groups like this are an absolute godsend to our finances that can allow us, by the way, to subsidize other groups and you could start to look at what would have to be true to be able to take the highest leverage changes possible. Um, because again, you've got this awareness system that is looking at all of the data, I mean the direction. All I want to do is run this company till I die and then have it outlive me. I'm really, really proud of what we're building and I think there's a good chance that I'll be able to do that. And if I can use technology and if I can pair it, because it does come back to the people with absolute world class leaders across functions which I think I have right now and will retain in the future, I'll, uh, be able to make my mom proud at the end of all this, I'll be able to say, I didn't become a doctor, but I was able to help fix some of what's broken. And, um. And that would be a life worth living to me. Andy Sack: And so what I guess I didn't in that beautiful sentiment. What I guess we're talk about the, you know, your eyes being bigger than your stomach as it relates to the AI journey in that. Nick Soman: First, I need AI to get its hooks in every function of the organization in a meaningful and transparent way. I need to be able to not say, hey, I've had some really good conversations with Julia about how we're going to be able to answer questions for members, but I need her and her function to have the aha moment. That has happened with me and with others on my team when AI takes this big chunk of work you thought you were going to have to spend your Saturday doing, and it says, no, it's done. And not only is it done, but I know how to do it better next time. So just feel free to ask again. I need that. Because what I've seen is that as soon as people have that, she's going to be coming to me with ideas. And one thing we do that's a little different at decent, we don't think of things as a global hierarchy. Now, obviously, the CEO lives in the United States, but I go to the people in the Philippines and I send them the same message. Look, I don't want to hear that you are the hands for the US Strategy operation. I don't look 00:25:00 Nick Soman: at things that way at all. You're the closest to the problems you're trying to solve. I need you coming to me with ideas. And if you aren't well equipped to do that, then you're not going to last on this team for very long. Because I'm going to hire the best people I can find in the Philippines. I'm going to hire the best people I can find in India. And I want to get everybody to have that aha moment where they go, forget what Nick wants. This is making my job and my life a lot easier because ultimately people are going to be more motivated by that than some stump speech that I can give. Then the second thing, and I'm just predicting in the future, as I was running through these five tasks for four different AI tools, knowing that there are another four I easily could have thrown into the mix, I thought, boy, this feels really inefficient. It seems sort of obvious to me that AI aggregators are going to be A thing as well. Just as it doesn't make sense to go to United and buy a plane ticket, you know, Kayak has evolved from that. I did such a good job policing itself and being able to identify the differences in its own thinking and figure out clarifying questions to get to a good answer. You know, obviously you got to figure out token spend because that's. This stuff gets expensive. But I do think ultimately, you know, we probably end up with a super intelligence that looks more like, you know, one AI to rule them and bind them because there's just not a great reason for me to be copying pasting things into to five different interfaces. Adam Brotman: And Nick, on the topic of, of, I'll call it data analysis with AI, so it sounds like in your companies, a lot of what you do, uh, in, in your corporate, collective corporate function is you, you have to do a bunch of data analysis or at least predict what something's going to cost and, and whatnot. Nick Soman: Actuarial underwriting. Yep, that's right. Adam Brotman: Yeah. And so did you before, before, you know, Chat gbt, did you, how did you do your um, analysis? And then have you been able to use ChatGPT for some of that analysis? Have you like, and, and do you like, is there like a concern around like pii? Do you like, do you obfuscate or anonymize the, the information before you run it through ChatGPT, even though you got Enterprise, which is the most secure version? Like, walk us through a little bit of like the data analysis flow and how it's helped you or where or where you have challenges with data analysis. Nick Soman: The way we did it used to be very dumb. And I have to tell the truth because Andy was there when we were doing it. Like, it took me way longer than it should have because we originally launched to serve self employed people, and when we launched to serve self employed people after like a year of doing it, we looked at our numbers and we're like, oh, here's the crazy thing. When we enrolled these people at the beginning of the year, which is during what's called the traditional open enrollment period, where your health plan might start in January, we did pretty good. They were relatively healthy. We were able to serve them at a low cost and give them a really good renewal. But for the people that enrolled in the middle of the year, we got our asses kicked. And then, you know, you start thinking about, well, why would that be? Oh, it's because they waited to get sick and then they said, well, now that I'm sick, I should probably get some health plan and these idiots are, are, you know, right here willing to offer it to me. That's the sort of thing that easily could have been identified as, you know, an early flag by sort of a well tuned AI system. Now, could it have been flagged by an expensive data team that's going to cost me a million dollars to hire in San Francisco? Sure, but what a waste of money. Uh, the opportunity is to get this thing looking at the data you always have to think about. Phi Pii is actually less of a concern because it's not connected to health. But we as an organization have sort of made that a priority from day one. And we are very careful to make sure that nobody's looking at data that doesn't need to be seeing it. Um, as far as I know, there are not yet laws about giving, you know, an LLM access to a whole lot of data and letting it turn itself loose. It's about what happens when you make it available to another human. So thankfully, at least for now, we, uh, have a fair amount of leeway there. But here's what's crazy. Like, I used to be a growth leader at gusto, and nobody wants to say this, but growth is really about how to connect all the crazy dots on the Internet to know as much about people as you can possibly know. And it turns out that's an incredibly relevant exercise to healthcare. That's an incredibly relevant exercise to making the best use of an LLM. Not to say it in a creepy way, but just as an example, if somebody gets on our plan and they have diabetes and we can know that up front, we can call them ahead of the plan starting, we can have a meeting with them the first day and say, this is your doctor. We pay for some extra stuff for you because we know that it's going to pay for itself in the long run. As long as you're willing to, um, to work with us, the more we can know, the better we can take care of people. And I guess it could be a little Big Brother if you wanted to look at it that way. But when I've explained it that way to our members, just, you know, tell us what you're willing to tell us and we're going to find out what we can and we're really going to translate this to getting you the best care you can. People don't seem to mind. Adam Brotman: Do you. Do you find that you give the AI systems, whether they be ChatGPT or any of the other ones, do you find that you give them just Raw data and be like, hey, this is the insights and the correlations 00:30:00 Adam Brotman: that we're looking for? Or do you tend to give them summaries and have them kind of work from the summary? You know what fidelity of data. Nick Soman: I like the raw data. I would just write because I want the raw data and so I would treat it the way that I'd want to be treated. Um, I like to throw it data because often it'll go, here's something you didn't notice or something that you wouldn't have of guided me in the right direction. It feels like, you know, I wouldn't buy a TI83 and then explain to it how to do simple math. Uh, I'm happy to give it everything. Adam Brotman: Nick, do you, do you find that you like when you have. Does it ever like chug on that data and you go, why is it having problems? And then you realize you need to clean up the data or you need to specify what columns are and things like that? Nick Soman: That definitely happens. Um, I've also learned that you need to break it down into constituent parts because especially the ones with a shorter context window, they just don't know what to do. Um, what I don't like, if I could complain is, and I didn't know that I didn't like this until I saw an alternative, but I think Gemini is the one that'll tell you, like, hey, the context window is not long enough for us to credibly answer your question the way that I know you're looking for it to answer. I thought of all the times that I wish ChatGPT would have told me that because it'll sometimes just start going, well, I don't know. So here's some shit I made up and it'd be nice to at least have a, you know, a flag on like level of competence or something like that. But I understand they made a different choice. Adam Brotman: Yeah. Andy Sack: Nick, I'm, I'm curious if there were. You had a magic wand and you could build a line of business AI app that, that increased your ebitda, uh, or was strategically important critical to decent. What are the top two line of business apps that come to your mind? Nick Soman: The first one really absolutely is I want AI to do as much of the day to day work to answer member questions as it can possibly do. And it needs to be extremely contextually aware. It needs to be privacy focused and it needs to not creep people out. And there probably needs to be a human in the loop. But like, let's minimize the impact and the effort of that human in the. Andy Sack: Loop, because this is not broadly be called customer support. Nick Soman: Is that like, yeah, that's a good way to think about it. Or customer service. Here's the deal with health plans and I used to work in consumers, so I can say this like, nobody wants to talk to me. Nobody wants to talk to a health plan. Nobody wants to dig into 17 pages of self serve paperwork, like answer my damn question. Let me move on with my life. If you need to tell me what to do, then tell me and make it really easy. And those are things that I think an AI is better equipped for than a human. And so that's the first thing. I really do think there's something to this idea of using everything that we know about someone to just divine what is the absolute best course of care. Now this comes back to philosophy and my parents and, you know, growing up with primary care doctors and understanding that the first thing is you got to give a shit about people. But if you give a shit about people, your tendency is to ask enough questions to actually make yourself useful. AI can answer and ask a whole lot of questions, can synthesize information without asking. The white whale for me is, you know, I got on this decent plan and it's not like any other plan that I've ever seen. You know, before the coverage even started, I got this email introducing me to Julia and you know, a really thoughtful and personalized course of care recommendations and some incentives to make me want to do them. Like, this is the direction we can go and the path to get there runs squarely through AI, not me. Throwing bodies at the problem and trying to do what UnitedHealthcare has done and build an army of low skill. Andy Sack: Yeah, excellent. Um, I told, I promised you that it'd be a moment for you to give a little pitch on decent. Here's your opportunity, Nick. Nick Soman: All right. Before any of this, I'm the only one in my family that's not a primary care doctor. And so I, uh, share this with you a little bit as bonafides and also for you to know what drives me. I watched my parents grow up and they went from talking about compassionate patient care to the point that my sister and I were like, please shut up, we don't want to hear about this anymore at the dinner table. It's not interesting. You know, she would of course go on to medical school. I watched them become disheartened. I watched them get basically chewed up and spit out by these big HMO insurance company entities and feel like thousands of doctors across the country are feeling. I didn't get into this business to dance for the insurance companies. I got into it to take care of people. And so I wish the model that they had found instead of becoming grouchy HMO executives in Seattle was this direct primary care model. Uh, it's very, very meaningful to me. There are 8,000 of these clinicians across the country that have really thrown off the shackles of a traditional insurance company. And I should say decent's not an insurance company. We're a third party administrator that enables self funded plans, uh, so that these primary care doctors can focus on what they love to do, which is delivering primary care. And what's really cool is we had a hypothesis seven years ago that now is definitively true, that if you give people enough 00:35:00 Nick Soman: primary care in the middle of their plan, you make it easy enough for them to access it. It actually feels like a concierge experience. But it saves the plan so much money because people get the primary care when they need it. They're kept healthy when they need it. They listen to their doctor and they do what the doctor tells them to do and it keeps them healthier that you can actually offer a drastically cheaper health plan. So 30 to 40% cheaper than these fully insured insurance company market rates. And, uh, it's really starting to work. And the neat thing that I wanted to share with you, Andy, and it's less about pitching who might be listening, it's like as a guy who loves you and really sees you as a mentor, like here's what's happening in my business. Not only are we growing and we're going to hit our number and that's, that's really exciting, but it's who we're growing with. The people that are in this direct primary care movement front and center are for the most part really choosing to get on our plans not just as partners, but as customers. And that's an incredible thing because you've got, I'll, uh, name some names, you know, Hint Health. They are a software company that powers the payments for 65% of the direct primary care space. They'll be a decent customer. Health Rosetta, uh, the guys that I used to look up to and go, God, maybe someday I'll be as smart as you. And I honestly still kind of feel that way. But they've helped us figure out how this plan should work. And then they liked it enough, they got it on it themselves. And there's five or six other examples like that that I just want to clear with them before I mention it. But you know, the beautiful thing is like the experts are picking us and most of the employers are looking to the experts. And so I think we're really well positioned to grow into the new year. Andy Sack: Adam, any, any closing comments or suggestions for Nick, Anything you want to highlight? Adam Brotman: Mhm. I mean, Nick, you're, I was, that was a great inspirational 45 minutes. Uh, I'm jazzed. I was up late last night, uh, Thanksgiving week, you know, craziness. And uh, you know, you, you've, you're inspiring me and usually uh, you know, people are looking for Andy and I to do that, but you, you gave a good shot in the arm, so hope to have you back. Uh, I mean, we could talk about example after example with you. So I, I, it would be fun to have you back on the pod. It was great. Nick Soman: A, let's do it and B, let's do it. When I put my money where my mouth is a little bit more and I can point you to some specific things that we've rol. That's coming. Andy Sack: Excellent. Well, Adam, uh, Nick, thanks. This concludes another episode of AI first with Adam and Andy. To stay updated on the latest in AI and access exclusive content, go to our website forum3.com, join the email list. Follow us on LinkedIn. We truly believe you can't overinvest in AI. Adam, that was a great conversation with Nick. Um, from decent. Let's uh, let's debrief, um, what stood out for you from the conversation? Adam Brotman: Well, the first thing that stood out was his enthusiasm and energy around the topic, which you and I can relate to. And it, I just, I just want to comment, I mean, commend Nick, but also just kind of comment on, you know, you know, we're at a stage of AI adoption in the enterprise where I think that that's the exception versus the rule. And I think we're going to see that change as we go into 2025. But that was cool to see him talk about the Iron man suit like, which is. You and I have talked about a limitless pill in the past or an Iron man suit like this idea where you sort of realize that this can help you across anything you're doing across any function. Yes, you can use it to automate and streamline things. But, but on top of that there's the inviting it to the table, ah, all day, every day point. Uh, that is probably the biggest opportunity for CEOs and leaders right now to instill across the organization and he's clearly doing that. And uh, gets it. And it was just, it's fun to see somebody who's in that mode. And I think that's a big part of the reason we're doing this podcast, is to get more of that in front of our listeners and our viewers so that we can. It's not just you and I talking hyperbolically about this technology, but getting into the real everyday use cases and showing like, look, this is what happens if you use it every day. And you know, we'll probably even, you know, you know, I think we pushed Nick on ROI a little bit and we've talked to our other customers about roi and we're going to see more real numbers come out in 2025. It was so new, but it was, it was cool and exciting and refreshing to see an AI first CEO like Nick, you know, on the show. Andy Sack: Yeah, I mean, I, I mean I've known Nick for a long time and so uh, I've seen him develop, I think what stood out for me from the conversation, there were three things that I think. So first, Nick is, I would say like a next generation CEO. He's an AI first CEO. He is turning himself and using not just one tool, but multiple large language models he says is working across 00:40:00 Andy Sack: three to five of them. So he's a, he's a next generation CEO, I think, um, uh, an AI first CEO, if you will. I think his business, the fact that his business, uh, he uses the, the mech suit, the Iron man suit for himself. But he also talks about how as a small company you can punch above your weight. And I think that's true of a big company that uses AI, they can punch above their weight. But this notion of both as an individual and as a small company using AI to punch above your weight, I think it's a great example of, of really small, using small companies that don't need to hire as quickly as they did previously and they can punch above the weight. And then third, I actually think that his business, his industry, um, which healthcare and helping um, direct, uh, providers really get a better customer experience through AI. Um, so I think the fact that he's operating in healthcare really stood out to me as an industry that's going to be transformed, um, and where AI really empowers, um, uh, a better customer experience both for the, the consumers of his product as well as the providers. Um, uh, so in every aspect I think the AI is really impacting the nature of his business. And so that really stood out. Those three things stood out to me. Adam Brotman: No, those are really Good points. The punching above your weight point, um, is. Is interesting because it gets to this, you know. You know, you and I have talked about the second aha, uh moment that can happen, which he's clearly had. Like, the first aha, um moment, which we sometimes call the holy moment, is where pretty much everyone's had it by now. Or at least not. Well, I shouldn't say that a lot of people had it by now where it's like you used AI, uh, Chat GPT or Gemini or whatever. Claude and you. And there's this for the. Everyone can remember that first time. They're like, holy, what is this? Like, how is it able to truly be something out of science fiction? Like, truly be like a smart human understanding me and being able to intelligently respond in a way that I never would have thought a software program could respond? That's like the first aha moment. Then. Then there's this. But then what happens is a lot of people see that and they're like, they got that because they asked it to write a job description. They asked it to help it write an email, um, or. Or maybe even kind of used it instead of Google and got like, a really thorough, Thorough, interesting answer. Um, which may or may not have been correct if you're using instead of Google. But the. But the point, I mean, uh, particularly if you're using it prior to, like, ChatGPT search being live. But then there's a second aha moment that happens, clearly, and Nick was talking about this, where you start to realize, wait a minute, this is like. This isn't like any of the software I've ever dealt with. It's more like having a BCG McKinsey consultant. It's more like having a data scientist. It's more like having an expert. It's more like having a chief strategy officer, uh, another marketer or whatever. So, like, once you start and it's like, kind of limitless and free, and that. That's when that second aha moment kicks in. And at that point, I think you start to, like Nick is saying, and I know we are infected with this. Like, you. When you have the second aha moment, you start to go, oh, I need to put everything I'm doing through this, like, immediately. Because it's gonna, like you said, it's gonna take. It's gonna uplift everything by a certain amount. And that's how you're punching above your weight. Because it's like, no, Nick can't afford bcg. I mean, I don't know that much. About his finances. You do, I think you're on this board. But the point is like I doubt he's gonna be able to hire a BCG or McKinsey, um, at his startup size. But he's got AI and if he knows how to use it effectively, like it's, it's. No, it's not the same but it's, it's kind of like having access to a top tier consultant and extra staff and extra experts and whatnot. And like of course that allows you to punch above your weight. So you're just constantly trying to figure out how to run as much of your business through it. Uh, and I, you can kind of feel that from Nick when you're asking him questions about it like that he's, he gets that. Andy Sack: Yeah, totally. I mean it, it's interesting. It dovetails with the conversations you and I have been having just generally about roi, which is that particularly as you're a small company having and you transition to operating much more closely with AI is this element where you actually don't have to build up the um, uh, cost base issue. Like, like he says you punch above your weight. Um, and so whereas mid sized companies come along and they're like, want to transition. It's like, oh, uh, what's the first thing, the first thing that most people go where they can 00:45:00 Andy Sack: in terms of a line of business where they can get an AI, it's generally they're either looking to increase revenue or reduce costs and it's easier to get the ROI clarity around a reduction in cost. And so I just think that that's interesting that as a startup you never, you don't have to do that as much because you never, never build that cost into the business. Adam Brotman: Totally. Yeah, you never, you never could afford it. So you were always just trying to figure out how to like get to growth without adding cost. Andy Sack: Right? Yeah, totally. Adam Brotman: Yeah. Andy Sack: I love this point, I love this point of uh, with his point that all the LLMs are underpriced. Yeah, um, that's, that's, that, that was great. And I think he's right. Um, I mean we've heard this from Paul Reutezer and others. Like, I mean, can you believe the level? Like here's a limitless, you know, uh, limitless, uh, intelligence as a service. Um, and they're only charging 20 bucks a month per person, per employee. It's unbelievable. Adam Brotman: It is unbelievable because if you think about it like there's. Yeah, Paul Racer makes a great point. He'll create like a great custom GPT that's really useful for his business amongst the 5,000 other use cases he and Mike and his team use AI for. But he'll be like, oh, that one GPT I'd pay, I'd pay $500 a month in other words, for that one call it self rolled feature that he made out of it with a saved prompt in the GPT. Like it's, it the ROI there is like 10x what he's paying for and that's one of a thousand things he's doing with it. So that's, yeah, the ROI point is, uh, M is, is uh, pretty obvious in that case. But, but by the way, I think that brings up a bigger topic that you and I should also talk about, which is that there, and I think Nick was mentioning this, which is just in general these systems are kind of underutilized. You know, it's interesting, I was, I was listening to CNBC this morning and Jim Cramer and David Faber were in, Carl Quintanilla or whatever were talking about AI and Jim M. Cramer was saying that he was using the free version of ChatGPT, I guess and uh, was having trouble getting it to do good research for him on companies because it was like getting stuff wrong and it was referring to like you know, old data. And I thought to myself, like I wonder if he's not getting it to use Search because that, and he's just asking questions that, that are uh, basically screwing up the fact that he's, he's not taking into account the training data cutoff date. But here's the thing that's not on him. Right? Right. You and I have seen this a lot of times. Like with technologies. We saw this with Blockchain, we've seen this with Metaverse, we've seen this time and time again. Like if it requires you to understand how the software works and whatever, like you're dead. Right. So that's not helpful. And yet here's this technology that is being used by hundreds of millions of people is probably the number one priority probably of most of the major hyperscalers and tech companies in the world. And yet it doesn't come with a user manual. You do have to have some um, AI literacy and proficiency to get the most out of it. And so like that's just a really interesting spot, right? Where like when you talk about things being underpriced, I think part of the reason is like a lot of CEOs and a lot of people still think of this as like a novelty. And it's interesting and kind of cool, but Jesus, it really worth $20 a month or whatever. And then there's other people that are like, oh yeah, oh yeah, I am turning my company. I can double the EBITDA of my company and the growth rate of my company with it. So it's worth hundreds. It could be potentially worth like tens of millions of dollars. And I was saying like, it's really interesting. I was listening to that CNBC moment where Jim Cramer was like saying, oh, you know, the free version of chat gbt, you know, it's giving me not great help on my company research and he's giving me old information. And I, I was listening to him and I thought, yeah, like probably because he wasn't, he, he wasn't thinking about the training data cut off as he was prompting. And then he may or may not have been able to take advantage of the search capabilities that are built in to ChatGPT4. Uh o. And then the next thought I had was like, well, geez, you know, there's no user manual. Like he hasn't been through a boot camp or something. So talk about the uneven distribution of understanding and usage. And so really that just goes. So it's fascinating. It goes to show like if you had some level of AI literacy, some level of AI proficiency and training, even minimal your ability to utilize this technology, the difference is between it being a novelty where you're not sure if it's worth $20 00:50:00 Adam Brotman: a month all the way up to like startup companies that are like using this as like a Iron man suit to create tens of millions of dollars of incremental enterprise value from the same technology for the same price. Like that's a really interesting like dynamic with where we're at with this technology right now where like it's that level of delta. Delta, which is interesting to me. Andy Sack: Yeah, I mean, I think that's interesting. It makes, uh. This is unrelated but related. What the, in, in the, in this. You're talking about the, the jagged frontier of, of, of people's individual usage. I want to come back to the point that I made at the beginning of what stood out to me, which is like really the healthcare industry and in particular healthcare insurance, which is decent business. Like if you just think about as an audience member, your own unique situation and your own unique, Your families. Everyone's got, I mean at least in my family, I've got a, uh, I've got a diabetic, uh, I'm a survivor of cancer. I have particular interests and needs, highly personalized information and data. And then you have these really complex regulated priced plans and it's always so hard and then you've got the web of providers. It's always so hard to figure out what plan to pick. And does that doctor, does this plan match me? Which doctor should I use? Am I like. And you want all that information served up. AI is the, is the single technology to actually make that user experience seamless. And it's kind of amazing because it's, that's very different than some of the other guests we've had. It's really a great, great example of high personalization and customization delighting the end user and really just how you know, I'm reminded of when we interviewed Reid Hoffman and he was talking to the Pope and the capability of having you know, an uh, AI on what was it, 8 billion phones worldwide. So that just stood out to me from the interview with, with Nick and it talks about the, to connect it to what you were saying. The jagged frontier. There's the, it's really an example of the, a positive example, the jagged frontier of you know, company fit and industry fit for the implementation of AI first experiences. Adam Brotman: Yeah, yeah, yeah, yeah. It really you just reminded me of, I can't remember who, who we heard this from. M. Nick maybe mentioned it or uh, alluded to it just now but, but I think we heard this, that AI is absolutely @ its best when you have large amounts of data. I think Nick was kind of referring to this large amount of data and not a lot of time to make a decision. In fact I've heard from another one of our influencers, I can't remember who it was saying if you want to know the best use case for AI, it's this generically stated concept of tons of information. Not a lot of time got to make a decision. AI is, that's where AI is pulling its weight the most because even if it's directional mostly right. It's better than you not having any help in making a decision. Right. And so can you start to think about like you're talking about healthcare, insurance, you know, financial industries. There's all these industries where like ah, there's all this data, like literal data, like unstructured huge amounts of data or structured data and you have to do analysis, make decisions like normally you have to have like bi teams and advanced analytics and like having an AI at your fingertips is like, it's incredible to think about. But then when you go back to like what constitutes data and you think about like that case study from our book where Matt Britton from Susie was saying, oh, I have all this data and I was not utilizing it. And my first thought was, like, thinking about typical data, like in a data warehouse or whatever. And, like, and he was talking about all the transcripts of all the sales and account calls. And I was like, oh, my God, that's right. Like, who's gonna. The sales manager is not gonna go back and read, uh, thousands of hours worth of transcripts, but AI could do that really easily in a matter of seconds and then basically give you all sorts of guidance and alerts and whatnot. So, um, to your point about healthcare and all this, like, personal information and the need to sort of sift through all these different variations of plan details, like, we're heading to a world where your own AI, your own ability to leverage AI as a worker and in a personal capacity is just going to be game changing. Andy Sack: I completely agree. Um, it was great to have you on Nick, even though he's not here. Uh, anything else you want to say about this before we wrap up? Adam Brotman: No, no. Great episode. Let's get some more Nicks on here. This is great stuff. We're learning a lot. Andy Sack: Yeah, I 00:55:00 Andy Sack: love the flag behind Nick. When he did the. Did the interview, it was funny. Okay, well, uh, with that, um, thanks for listening to AI first with Adam and Andy. Uh, stay updated on the latest in AI and access exclusive content. Visit forum3.com follow us on LinkedIn. Sign up for our email list. We truly believe you can't over invest in AI learning. Onward. 00:55:29