Chapter CEO Cobi Blumenfeld-Gantz - medicare complexity, $100B fraud and how tech can fix it

Cobi Blumenfeld-Gantz is the co-founder and CEO of Chapter, an AI-powered Medicare navigation platform valued at $1.5 billion that helps seniors navigate coverage and enrollment. In this episode of World of DaaS, Cobi and Auren discuss: Why medicare fraud costs over $100 billion annuallyHow 24,000 medicare plans create impossible choices for seniorsWhy government tech keeps failing with Accenture contractsProvider network data accuracy and AI solutionsYou can find Auren Hoffman on X at @auren...

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:02Auren Hoffman?

  Hello, fellow data nerds. Welcome to World of DAS. I'm your host, Orin Hoffman, CEO of Incubate and GP of Flex Capital. Discover more episodes, get weekly data as a service news, original content, articles on data and more at worldofdas.com. That's worldofdas, D-A-A-S, dot com. Hello, fellow data nerds. My guest today is Kobe Blumenfeld-Gantz. Kobe is the co-founder and CEO of Chapter, an AI-enabled Medicare advisory platform that builds comprehensive data and models to help seniors navigate their coverage. Chapter raised a Series D in April 2025 with the company valued at over $1.5 billion. Prior to founding Chapter,

• 0:46Auren Hoffman?

  Kobe worked on the US government team at Palantir. Kobe, welcome to World of DAS.

• 0:51Speaker 01

  Thanks so much for having me.

• 0:53Auren Hoffman?

  I'm super excited. Now, Medicare might be the most complicated marketplace in the world, certainly that I know of. What is the core misunderstanding that people have about how the system actually works?

• 1:05Speaker 01

  There's so much. I think there's a few big things. One is that most people assume that Medicare is just automatic, that when you turn 65 or retire, you just get it. But that's not all what happens. You actually have to take a lot of actions. You have to sign up for original Medicare and then supplemental coverage because the government offering only covers up to 80% of expected health costs. So almost everyone gets supplemental coverage. There's a lot of decisions. And then because of the way the risk pools work, if you sign up at the wrong time or you delay signing up or you miss your window, the government imposes lifetime penalties on you.

• 1:38Auren Hoffman?

  For what?

• 1:39Speaker 01

  Yeah, it's actually really messed up. My mom has this. So if you miss your signup window, the government will add-

• 1:46Auren Hoffman?

  Like if you sign up at 66 instead of 65 or something?

• 1:50Speaker 01

  You have a few months of a window to sign up for, but if you miss that few-month window, usually it's seven months, sometimes it's eight months, depending on your situation. But if you miss that few-month window, then the government will start adding roughly 1% of premium every month. So if you sign up 12 months late, you'll be paying a 12% higher premium every year for the rest of your life.

• 2:09Auren Hoffman?

  What's the reasoning for that? What's the rationale? You would think it's good for the government if you sign up late because then they don't have to cover you for that period of time, right?

• 2:16Speaker 01

  The policy rationale actually does make sense. It's because they need to create a risk pool and they can't just have people waiting until they're older and sicker. And so when you're 65, you're actually young for a Medicare age and they need to get the young people and healthier people into the risk pool, paying their premiums earlier. Otherwise, the whole system breaks.

• 2:33Auren Hoffman?

  Oh, okay. Because for Medicare, you still pay something, even though it's covered. It's a good deal, but you still have to pay. And so that money that goes in essentially, when you're 65, they're making money off you, but by the time you're 75, they're losing money on you.

• 2:47Speaker 01

  Yeah, directionally, yes. Because you pay Medicare part A, payroll taxes your entire life. So you've been contributing to the system. But there's another part of Medicare called part B, which you have to pay for each month. And then there's all these other parts exceedingly complicated.

• 3:01Auren Hoffman?

  Okay, interesting. But that 1% more per month, that sounds like a lot, actually.

• 3:07Speaker 01

  Yeah.

• 3:07Auren Hoffman?

  That sounds like a pretty high. I didn't realize it was that much higher.

• 3:10Speaker 01

  Yeah, and it can get up there. Again, there's a reasonable rationale for it, but the government makes it so complicated and so opaque to know when you're supposed to do what, that it's actually very hard.

• 3:21Auren Hoffman?

  It's not like you just get an app reminder being like, hey, you haven't done this yet, or something like a call from a friendly person.

• 3:28Speaker 01

  If you're a chapter member, you do. But no, if you're just a citizen and the government will mail you a 350-page book that's called Medicare and You.

• 3:37Auren Hoffman?

  Oh, my gosh.

• 3:38Speaker 01

  That they expect people to read. And then, yeah, you're supposed to parse that.

• 3:41Auren Hoffman?

  Oh, wow. Okay. And what happens if you move between states? So let's say you move from Connecticut to New York or somewhere like that. Do you have to re-sign up again or something?

• 3:50Speaker 01

  Yes. So the original Medicare, that first part, is federal. So you don't have to do anything there. But for the supplemental coverage, which almost everyone has and has to deal with, that is either county or state-based. So you will have to likely switch your plan at that point.

• 4:04Auren Hoffman?

  With counties, if I move counties, I have to do it?

• 4:07Speaker 01

  If you're on a Medicare Advantage plan, this is getting really in the weeds. But yes, if you're on one type of Medicare plan called a Medicare Advantage plan, it's at the county level. If you're on another type of Medicare plan called a Medicare Supplement Plan, it's generally at the state level, though sometimes they parse the state into like two or three zones.

• 4:22Auren Hoffman?

  I hear of people, so what about Snowbird? They kind of live in Michigan, but then they're also in Florida. Do you just kick the place you're more likely to be in or something? Or you pick the place that's somewhat cheaper? Or how's it work?

• 4:34Speaker 01

  You have to go with your primary residence. So whichever state your primary residence is, you have to get a plan that covers there. But that's one of the reasons.

• 4:41Auren Hoffman?

  So for most people, that would be Florida, I would presume that. And if those two things, yeah, yeah, okay.

• 4:46Speaker 01

  And so you pick that. And then for people who travel a lot, they will probably go on to what's called a Medicare Supplement Plan because it does have wider national coverage, whereas Medicare Advantage Plans tend to have more narrow networks that are regional.

• 4:59Auren Hoffman?

  Okay, interesting. And you hear of sometimes these scenarios where people are on two plans. They're on a plan in Connecticut and a plan in New York. And somehow these insurance companies are like double billing Medicare for these people. Is that as common as you would read about in the newspaper?

• 5:15Speaker 01

  Well, that's fraud. I think that is not super common that someone's actually on multiple plans. There is rampant fraud, though, but it manifests in a slightly different way. Or there are a lot of ways it manifests. There's fraud at the provider level, which is the doctor is billing the insurance carrier or the government for far more than they should be. That might be, I'm billing as a doctor for an hour-long appointment that has a lot of intense tests when it was actually a five-minute conversation that was very routine.

• 5:43Auren Hoffman?

  From what I understand, that's close to 100% if doctors are doing some upcoding here and there or something like that. It's extremely hard to catch, especially if it's, they're only padding their fees by 20, 30% or something.

• 5:55Speaker 01

  Yeah, and it's hard to say how much of that is fraud. Yes, most doctors are charging probably more than they need to in every case, but most doctors, I don't think, are actually defrauding the government. But there are a small number of doctors who account for the vast majority of the fraud. That's one big area. And then it shows up in other places, too, where they'll basically, durable medical equipment, DME, is one of the largest categories of fraud, where people will prescribe a wheelchair or these other like larger equipment items that people don't even realize that they're being prescribed.

• 6:23Auren Hoffman?

  Yeah, I know people like have gotten elbow braces sent to them, which they don't even need. They never asked for, and it's sitting in their closet. And then they found out later that somebody billed Medicare $2,400 for it or something like that.

• 6:37Speaker 01

  Yeah, maybe closer to $5,000, but yeah.

• 6:39Auren Hoffman?

  What is the magnitude of the fraud? Is the fraud, let's say Medicare or Medicaid is a trillion dollars a year? Is the fraud 100 billion a year? Or what's the magnitude of that fraud?

• 6:51Speaker 01

  We don't know exactly. Estimates are probably on the order of 10 to 15% of all spend. So it's quite material.

• 6:57Auren Hoffman?

  So we're talking potentially over 100 billion and?

• 7:00Speaker 01

  Yes, yes, for sure. Over 100 billion dollars.

• 7:01Auren Hoffman?

  Okay.

• 7:02Speaker 01

  For sure.

• 7:02Auren Hoffman?

  Once you get to 100 billion, that's real money.

• 7:04Speaker 01

  Yeah, yeah, no, it's real money.

• 7:05Auren Hoffman?

  Even for the government, that's real money.

• 7:07Speaker 01

  It's real money for the government and it also creates all these downstream implications that are really pernicious for the entire health system because it destroys trust in the system. It makes everyone upset and it can actually lead to worse health outcomes.

• 7:20Auren Hoffman?

  Raises costs.

• 7:20Speaker 01

  And it raises costs. Yeah. So it actually has a magnified effect beyond just the direct costs.

• 7:25Auren Hoffman?

  That's more than half of the army's budget. So it's real dollars. We're talking about real, real, real money here.

• 7:30Speaker 01

  Yeah, people assume that the Department of War now has the largest budget. HHS, the Department of Health and Human Services, of which Centers for Medicare and Medicaid Services, CMS is a sub-agency, has a budget that's 2X larger than our defense budget.

• 7:43Auren Hoffman?

  Well, and of course, that's not including the VA, which also is in that same healthcare world as well. A lot of people are committing fraud against the VA as well. We've seen technology try to transform. Well, certainly it's transformed retail. It's certainly transformed travel. It's starting to transform banking. Why is it not in that Medicare world yet? Or maybe it is and I don't know about it.

• 8:05Speaker 01

  I think there's a couple of reasons. One is it's a problem that you don't really face until you are in your 60s and you probably aren't aware of it unless you have parents who are going through it and you're in contact with them about it. And so because most technologists are younger, the world tends to work, it's not a problem that they feel. And so most companies build for what they see as problems and they don't see these as problems. So that's one big reason. A second big reason is the regulatory regime. There it is very hard to access this market and it's a very messy market. When I was starting the company, I had so many people, insurance carrier execs, insurance brokerage execs, tell me that what I was doing was either illegal or impossible or

• 8:47Speaker 01

  both because they just have pure path dependence and don't think about anything. And so there is real entrenched interests and stakeholders here that make it really hard to enter.

• 8:57Auren Hoffman?

  And some of these things that people have been wanting to do for a while, these aren't partisan things. These are just good government things. I assume everyone wants these to happen, wants these things to get better. Why is it so hard to change?

• 9:10Speaker 01

  There's a couple of aspects. There's the technology aspect and then there's the policy aspect. On the policy or regulatory aspect, there are so many different stakeholders. There's the insurance carriers who have one set of incentives and very powerful lobby. There's the doctors who have another set of incentives and quite powerful lobby. There are lobby groups like AARP that are in theory advocating for seniors directly. You have politicians who want to bring jobs to their states and want to claim credit for advocating for seniors because seniors tend to vote more. So you have all these interests and they're often in conflict, not always, but often in conflict. And so that makes it really hard to make real progress. Another element is that the Medicare regime has been Frankenstein together since 1965 and more and more regulations over time have been added for different parts.

• 9:55Speaker 01

  And it's just so complicated and there's so much in it that it becomes very difficult for people to move it and make changes.

• 10:03Auren Hoffman?

  Is it the US healthcare system itself is just completely Frankenstein and messed up? And then you're building a whole other system on top of that Frankenstein system?

• 10:13Speaker 01

  Exactly.

• 10:14Auren Hoffman?

  It's like a compounding Frankenstein?

• 10:16Speaker 01

  It's like turtles all the way down if the turtles had been built by Dr. Frankenstein. Yeah.

• 10:22Auren Hoffman?

  Okay.

• 10:23Speaker 01

  Exactly.

• 10:24Auren Hoffman?

  Okay.

• 10:25Speaker 01

  And then on the technology side, I actually think in some ways the technology side is easier to solve and would help a lot. But right now, and we can talk about this later or now if you want, but the way that HHS and CMS specifically are addressing technology is just they're repeating the same mistakes that every administration has made for the past 25 years or 30 years. In fact, it's the same Obama era people who were involved in the healthcare.gov debacle running most of the technology. It's just ridiculous.

• 10:52Auren Hoffman?

  And the problem is there's, they're saying Accentures, let's spend $30 billion creating this weird system that will never work? Or is it some other typical government problem? Or is there some other thing that they're doing?

• 11:06Speaker 01

  It's a pretty similar flavor to that, to the typical government tech acquisition problem, but it's a little bit different with HHS specifically right now, which is at least at CMS, they want to build their own data infrastructure and they have the right goals. The goal is to enable the private sector to build on top and augment the government, which I think is the right goal. But the way that they're doing it is they're trying to build the data infrastructure, which you and I know is actually the hardest thing to build and to build well. And so the fact that they want to build it in-house is just silly. The way that they're trying to build it in-house, aside from like, that's a bad idea. But the way that they're trying to build it in-house is they're relying on the Accentures and the Deloittes and the Booz Alans to build something that'll never exist.

• 11:44Auren Hoffman?

  Yeah. Once you get Accenture and Booz Allen in the world, the chance that'll work is basically zero. It almost never works. And then, of course, they're going to charge you gazillion dollars for it. It's always bad. You can't see at any reasonable cost to ever be good. And by the way, it's not their fault. It's not always Accenture's fault because they have all these weird requirements. They get project management work. All of a sudden someone goes and imagine you're constructing a building and all of a sudden in the middle of it, they're like, oh yeah, it's got to be four stories bigger and it's got to have a high-speed elevator that goes through the middle and not on the side. And you're like, wait, what the heck? So it becomes really, really hard to go do all these things.

• 12:21Speaker 01

  Yeah. I mean, working with the government is hard. A lot of requirements, a lot of challenges.

• 12:25Auren Hoffman?

  We have a new regime. We have Dr. Oz who's in there now. We have these other, they seem to at least be saying all the right things from like, okay, we want to share more data, we want to move things faster. Is there a way from like the top down to do this stuff if they want to?

• 12:40Speaker 01

  It's a good question. Dr. Oz, I think, is fantastic. He's super smart. As I said, he definitely has the right goals. Empower the private sector to build things, as he likes to say, CMS is open for business for the private sector, which is awesome. There's a couple of hard parts right now. One is defining the right goals within that, what do we want to go tackle? And then how do we do it? And they are getting a lot of pushback from the careers, from the civil servants who have been there for decades. And the pretty typical trope of a civil servant is, great, you're here, you want to change things up. I hear this every four years. I'm going to be here a lot longer than you are. I'll just slow things down and I'll outlast you. As much as the politicals and the appointees are aware of that and want to change it, it is very hard to change despite motivation. So that's one aspect.

• 13:23Speaker 01

  But I do think that they are making some of the same mistakes that many of past administrations have made, which is really not bringing in what I'd say is good technology. They're really relying on the old-school Accentures, Deloittes for reasons that despite them saying they know that that's not right, they still continue to do it. And at least from what I can tell, a lot of it is actually they aren't aware. They're deputizing to other people in their orbit, hey, go do this thing and do it quickly. And then those people are just relying on what they know, which is, okay, I'm going to go ask Accenture for how to do this thing.

• 13:53Auren Hoffman?

  And spend a gazillion dollars.

• 13:55Speaker 01

  Yeah, and spend a gazillion dollars, ask for more budget. And it's hard because from the very high level, is Secretary Kennedy going to care about what tech is used at CMS or FDA? Not really, and he probably shouldn't, or he shouldn't have to. But it's just one of these things where it's so entrenched where unless I think the secretary or someone at the DEPSEC level or undersecretary level gets involved, I'm not sure that much is going to change.

• 14:18Auren Hoffman?

  Yeah, I remember when the VA tried to do this, and I think there are something like 30 or 40 billion into this whole transformation that they've been trying to do, and it hasn't really worked at all. What can we have learned from that debacle?

• 14:33Speaker 01

  I mean, I have a heuristic, which is anytime any organization mentions the word tech transformation or modernization or innovation, it's already failed.

• 14:42Auren Hoffman?

  Correct.

• 14:44Auren Hoffman?

  And also, anyone who's the VP of innovation is always the stupidest person. That's why they got that title, so just get rid of that person immediately.

• 14:52Speaker 01

  I was speaking at like an old school healthcare panel a couple of years ago with a bunch of hospital execs. And I was the only non-hospital exec on the panel. And they were all talking about their innovation arms and how much progress they've made. And I was just like, guys, I don't mean to be mean here, but once you have an innovation arm in your organization, it means you've already failed. If you have a dedicated unit that's responsible for innovation, you actually need the entire company or organization to be moving things forward, not one group of people.

• 15:20Auren Hoffman?

  I've been involved with some of these companies before. These companies are very strategic. They'll have like a sales department, they have an operations department, they have all these things that are really important for the health of the company. And then they have someone who's not really performing, and then they put them in the innovation department because, well, why not? You just put them there, it's a nicer path than doing something else with them. And they can't really do harm in that, and they go to their conferences and they speak on the panels and stuff like that.

• 15:44Speaker 01

  That's how I wound up on the panel with them, exactly.

• 15:46Auren Hoffman?

  Yeah, yeah, exactly. Whenever you see that title, you're like. Okay, get away. And by the way, this is just a quick side for anyone in startup listening. Never hire somebody who had a innovation title at a big company or any company.

• 15:59Speaker 01

  And I also think it's an easy trap for people, for especially for startup founders, building for the government or working with the government to try to work with the innovation units in the government. So you have science and technology divisions in most agencies. Those are almost always where programs go to die.

• 16:14Auren Hoffman?

  Well, sometimes those are good. They're actually in the government. There are actually some good people, I think, in the innovation units. I just don't think that's true in the for-profit. Because the government, they're so sick of how things are going. They're like, I just got to get to a startup-y place in the government to try to make something happen. And so sometimes you have these rebels that show up that actually get stuff done.

• 16:31Speaker 01

  The people I think are really good. I think it's very hard, though, to move from a science and technology contract to a non-science and technology contract.

• 16:39Auren Hoffman?

  That's probably true.

• 16:40Speaker 01

  But I agree, the people are very good.

• 16:41Auren Hoffman?

  Yeah, that's a really good point. What is some sort of technology problem that somebody like me would not understand? That it just gets in the weeds of moving data and the health world?

• 16:53Speaker 01

  So, taking a step back, we started the company really focused on Medicare. We're now building a lot of AI and technology for seniors, really because we've built so much trust with our members that we want to help them with more aspects of their retired lives. We want to build the new platform for retirement across their health, their wealth, and their time. So within health specifically, I don't know if you're familiar with Siegel's Law?

• 17:16Auren Hoffman?

  No.

• 17:16Speaker 01

  Basically, if you give someone one watch, they'll know what time it is. If you give someone two watches, they'll never know what time it is. And there's a pretty common analog in healthcare, which is provider networks. The very simple problem of how do you know what doctor is in network with which plan? So I'm on a Medicare plan or I'm on a health insurance plan. How do I know what doctors are covered? Sounds simple. It's actually a very complicated problem. Many companies have been built just to solve this and many have failed. We have actually been able to solve this at 97% accuracy, which is far higher than anyone else. And one of the hard parts here is if you just rely on one set of provider networks or one data vendor for provider networks, if you have one watch, you assume it's right. Turns out it's not right.

• 17:58Speaker 01

  And so what we do is we integrate a lot of different provider network data sources. We have ML models running to know, given a lot of our experience, because we enroll so many people at this point, that we know which data providers.

• 18:11Auren Hoffman?

  Can't you just have an AI call them and just be like, hey, do you take a blue shield in Missouri for this particular thing?

• 18:20Speaker 01

  Yeah, if the doctors knew the right answer.

• 18:22Auren Hoffman?

  Okay, got it. So the person answering the phone, which isn't usually the doctor, may not even know.

• 18:28Speaker 01

  Exactly. So we have these ML models that weight different sources based on a lot of criteria for the given request. Then on top of that, when we're still not sure, we do have a headless AI agent that goes out and checks the insurance carrier website live, authenticated to the user, which is cool, so we actually know that it's the right user, pulling the data back and use that as an additional input. And now we're starting to roll out some voice AI that automates other parts of that. But the ultimate question is what's actually the source of truth? Because there isn't one, which is scary. So we've been able to get it to pretty high accuracy through a lot of these heuristics, but you have to take a lot of these data sources because there's no one data source that has over really 50 or 55% accuracy, even the carrier itself.

• 19:08Auren Hoffman?

  Let's say I'm a senior, I go in and I get some sort of test. The test comes back inconclusive and they say, hey, you should go get a biopsy or something like that. Getting a biopsy for anybody, especially seniors, is not fun. It's a terrible thing to go have to do. And I assume they just tell that to 99% of people to go get one, where maybe only, maybe they should only be saying that to 20% of people. Is there some sort of smart tech we could throw in there? I know it's not necessarily in the doctor's interest because they get money for doing the biopsy, but is there something that would make the seniors' life better there?

• 19:44Speaker 01

  Yeah, I think it has to be at the patient level because as you said, the providers aren't incentivized to do that and it's not entirely their fault. The entire medical education system is really heavily focused on these intense diagnostics. Sometimes are appropriate, but often aren't, as you said. And so I think we need to empower the patients, seniors and others, to know that, hey, given the data I have available, maybe I should go get a second opinion or maybe I should try these other things first. And it's hard to do that in a way because we certainly as Chapter and any AI provider is probably not a doctor. They're probably not going to have a medical license, at least not today. So how do you do that without stepping over those lines? I think it goes to the patient empowerment.

• 20:25Auren Hoffman?

  And simply it could be like, okay, I want to easily provide all my scans and all my history and all my whatever to somebody else who can very quickly, and maybe even doesn't even have to meet me in person and maybe even is based in Alaska or something, but can very quickly review this and have some sort of, and maybe if three doctors all say the same thing, okay, that seems reasonable. If they start to disagree, well, maybe I should dive into it more.

• 20:50Speaker 01

  Yeah, I think that type of thing is very doable. It really just comes down to the trust. Who does the patient trust? And especially seniors, they tend to trust their primary doctor that they've been seeing for a long time.

• 21:01Auren Hoffman?

  Which makes sense. I mean, that's probably who they should trust.

• 21:04Speaker 01

  Yeah, it makes sense, but it's those small nudges to get people to realize that. I like to use the gas station attendant analog here of 25 years ago before Google Maps and even before MapQuest, we would ask the gas station attendant for directions because they know the local market and you trust them because they're there. But they don't actually know that much. Whereas today, we trust Google Maps to tell us what's the right direction. But that takes time and that takes a lot of like repeat behavior.

• 21:29Auren Hoffman?

  Okay, that makes a lot of sense. What other data problems that exist in this world?

• 21:35Speaker 01

  So many. There is no single list of Medicare plans. It's the simplest thing. Where do you find a list of all Medicare plans? You actually have to integrate data from a lot of different places.

• 21:45Auren Hoffman?

  How many Medicare plans are there roughly?

• 21:47Speaker 01

  Roughly 24,000.

• 21:49Auren Hoffman?

  Geez, that's a lot.

• 21:52Speaker 01

  Yeah, it's at the county level, so there's a lot of them.

• 21:54Auren Hoffman?

  Okay, right, so there's 3,000 counties in America.

• 21:57Speaker 01

  It's not that crazy when you think about it like that.

• 21:59Auren Hoffman?

  How many Medicare plans would I have? Let's say I'm a 66-year-old man. How many plans would I roughly have to choose from?

• 22:08Speaker 01

  Roughly 100.

• 22:09Auren Hoffman?

  Oh, that is a lot.

• 22:10Speaker 01

  And no HR department to help you.

• 22:12Auren Hoffman?

  Wow, 100, holy. Why so many? Is because there's a gold-plated plan and there's a plan that I can go to with my furry dog. Why so many of these things?

• 22:22Speaker 01

  It's a competitive market. The Medicare Advantage market is quite competitive. So there are plans that have different networks, different price points. There's the gold, silver. They're not called this in the Medicare market. They are in the ACA market, silver, gold, platinum, elite, whatever. But there's lots of that. And then you also have this other type of Medicare, Medicare Supplement, which has another dozen or so plan options for you in that.

• 22:43Auren Hoffman?

  Oh, so there's like nested plans. I could choose this one and then based on this one, I could choose this other one. And it just doesn't seem like Medicare is that friendly to seniors. Seniors seem to like Medicare, they vote for Medicare. But based on what you're saying and based on the people I know who are on it, it doesn't really seem like it's a good thing. It seems like it could be so much better if we really want to take care of our seniors.

• 23:05Speaker 01

  Yeah, the core program, the actual coverage that you get is quite good. It's all the things around the edges, the access, the technology, the navigation. That's really bad, to your point.

• 23:17Auren Hoffman?

  Right, you get good care, you get good healthcare, but it's all these bureaucratic burdens that seems like something that we could solve.

• 23:25Speaker 01

  Yes, I mean, that's what we're trying to solve. But yes, absolutely, it's totally solvable.

• 23:29Auren Hoffman?

  It's like going to DMV and ultimately, yeah, you get this cool, nice thing that's called your driver's license. But you want to have a good experience when you go there.

• 23:36Speaker 01

  And there's a lot to do to fix it.

• 23:38Auren Hoffman?

  If we wanted to make CMS way more friendly to ecosystem, to innovators, to tech companies.

• 23:47Auren Hoffman?

  To have more competition in there. What could we do to make that happen?

• 23:52Speaker 01

  One big thing, I think CMS should adopt and really HHS writ large should adopt actually a similar model to what the Defense Department has adopted over the past decade, which is much more openness to newer primes. Like you have the Andurals and the Palantirs of the world that have done very well in DOD, but the same thing doesn't exist in HHS.

• 24:13Auren Hoffman?

  It's a good point. All right. So we were talking about in general Medicare for seniors that it's just so bureaucratic and so hard for them to know what to do. And I really feel like we're not taking care of the people in a way that we should. We owe it to the seniors to give them a great experience. Is there something else that we can do to help them?

• 24:34Speaker 01

  So I think the government could certainly open opportunities for smaller companies that are more advanced to help constituents more, to help Medicare beneficiaries more. That's one big area. I think the Department of Defense has done a good job opening opportunities for what used to be smaller companies like Palantir and Anduril that are now quite large. But the same thing hasn't existed at HHS or CMS where there aren't really opportunities for small companies to take on big meaty problems. There are lots of small, as we were talking about, like innovation armed things where small companies are part of it. But what I actually think HHS and CMS need is they need to give smaller companies that have proved themselves, not tiny companies, but smaller companies, the opportunity to take on these big meaty problems and be primes.

• 25:17Speaker 01

  And then they can have the technical capability and the knowledge to subcontract out where they need to.

• 25:22Auren Hoffman?

  So I know in the Defense Department, there are these five quote unquote primes, and then there's new primes like Palantir and Anduril that are coming in. In the CMS world, how many quote unquote equivalent companies are primes?

• 25:37Speaker 01

  It's a good question. I don't know exactly. It's probably about the same number. It's probably like five to six.

• 25:41Auren Hoffman?

  Okay, got it.

• 25:42Speaker 01

  It's the same five to six companies that control most of the budget.

• 25:45Auren Hoffman?

  And how does Epic play with all this stuff?

• 25:48Speaker 01

  So Epix, most people don't realize that Epix's primary product, obviously, is an EMR and HR. But the primary reason they're so successful is actually because they control billing. People think it's just patient data. It's not. It's actually the billing that is really where their core power lies.

• 26:03Auren Hoffman?

  So it's like the SAP for hospital systems or something?

• 26:07Speaker 01

  Yes.

• 26:07Auren Hoffman?

  Oh, so it's like a general ledger even in there.

• 26:10Speaker 01

  And that's part of why they have such ridiculous lock-in, because it's so hard to switch off your billing system.

• 26:16Auren Hoffman?

  Got it. It's like a monster company that is still private, right? It's like 100% owned by one person, right?

• 26:23Speaker 01

  There are a couple of these companies. Another one in more of the defense space is Esri. I don't know if you know them. They make a product called ArcGIS.

• 26:30Auren Hoffman?

  I know them very well. Yeah. Jack, who was the founder and CEO, it was the very first guest on the podcast. Now we're moving to Kobe, but we had Jack as our first guest. Yeah.

• 26:40Speaker 01

  One day, one day. But yeah, there are these companies that I know very little about Epic personally and directly. But it does seem like they have this interesting mystique, almost like Palantir used to have, where not that many people know exactly what's going on. They're all on a campus in the middle of nowhere. I have no idea what's actually going on and if it's in the same ilk as Palantir, but it's interesting to look at.

• 26:59Auren Hoffman?

  Yeah. And the campus is beautiful, by the way. It's like a gorgeous place to work and it's super fun and quirky. And you said, when you're at Palantir, one of my friends, Stefan Cohen, I think he interviewed thousands of people at Palantir. What did you learn from that experience about signals of people?

• 27:17Speaker 01

  Stefan is probably one of the greatest assessors of talent in the world, and there's a lot to learn there.

• 27:23Auren Hoffman?

  What makes him so good? Is it just reps or has he figured something else out?

• 27:28Speaker 01

  My impression is the reps have given him such strong intuition where he can pick up on things that aren't rational. He wouldn't be able to say like, oh, this makes sense why this signal is signal. But he's been able to create that, I think, just through a lot of reps. Like, most of his interviews are nine or 10 minutes long. He's not getting into depth on anything.

• 27:48Auren Hoffman?

  So what can you do nine or 10 minutes? I guess you could see how someone presents themselves. You can just ask them like a super hard question. Or do you be like, hey, how you do it? Like, what does one do during that time?

• 27:58Speaker 01

  What he normally does is, I don't know if it's intentional, but he often just tries to, I don't want to say scare people, but just put people off kilter because he has such an unconventional approach to interviewing. He'll be lying down on his couch when you walk in and then just start talking to you from there. So it's, it's just off putting. And then I think he likes to see how you react to that. But from my time, I find that the most effective ways to get signal are to ask three or four levels deep. There's no one question that gives you all the signal, but you learn so much by asking the fourth why, the fourth how, or go deep. Tell me more, tell me more, tell me more.

• 28:32Auren Hoffman?

  And it's something that they obviously know something about.

• 28:35Speaker 01

  Yes, or it should be.

• 28:36Auren Hoffman?

  So you did this at this last company, or you wrote this thesis about underwater farming and you just go deep on it?

• 28:43Speaker 01

  Exactly. I take whatever they've brought up as something they should be an expert in. And then I make up questions on the fly about it. So they can't give, oh, I don't have domain expertise as an excuse, and it's also just more fair. And then I ask a lot of questions about that thing to see how much they really understand it, how they react when they don't know something. I don't want people who are lying. I want people who are just honest and open, what they know and what they don't. And you can get a lot of signal that way.

• 29:07Auren Hoffman?

  I find that's an interesting, especially if someone wrote like a long thing about something, or they built some big project, really diving in deep. Is, at the very least, as an interviewer, I learn a lot.

• 29:20Speaker 01

  Yeah.

• 29:20Auren Hoffman?

  So at the very least, I spent my time in a productive way. Cause that's also something too, like, okay, what did you get out of this as the interviewer? Yeah, you want to assess the talent, but hopefully both of you guys came away from the meeting, learning something.

• 29:33Speaker 01

  Exactly. That's always a great sign when you learn something as the interviewer. I have some nits or some very small things that I've picked up that are actually, they're not necessarily PC, but they're definitely good predictors of whether someone's probably good. One is the language they use. If someone says the words passionate about, they're just probably not good. I know it sounds really weird, but it's true.

• 29:58Auren Hoffman?

  I'm thinking back, you might be right. I'd never thought about this before.

• 30:02Speaker 01

  I've interviewed far fewer people than Stefan.

• 30:04Auren Hoffman?

  Of course, hopefully everyone will now know this because we have a billion people that listen to this podcast. So interesting. What are some other like these little tells essentially?

• 30:14Speaker 01

  I think grit goes a long way, especially in a startup. The reality is if you're smart and hungry, most things aren't all that hard to learn, especially with AI, with access to technology and information we have. So I find that people who will really lean into problems, really think through problems live instead of backing away from them is a really good sign that people can be good. And that manifests in a lot of different ways. And then, I mean, I'm also just pretty open. I ask, are they looking to coast? Do they want to go work at Google? That's great if they do. Do they want to grind? How much do they want to grind? And I try to just have the conversation.

• 30:47Auren Hoffman?

  I assume that question's already being asked before you mute them. Or there are some sort of written interview or written questionnaires like, hey, are you okay with working long hours or something?

• 30:57Speaker 01

  Yeah, but I think it's different when the CEO asks because people, people actually answer it a little differently. So it's sometimes helpful to repeat questions. And then the last one that I find is a very helpful signal is I ask, just to walk me through every major career decision you've made. Why did you go to this college? Why did you take your first job? Why did you leave your first job? Why did you go to your second job? And you can get so much signal because you can tell if people are status chasing, if they're money motivated, if they're interested in certain types of problems. Are they going to be at your company for six months? And then like, you can get so much info from that.

• 31:29Auren Hoffman?

  And you can even ask them in retrospect, was that a good decision or something like that? You went and got an MBA. Was that a good decision to go do that? And it's always helpful to have someone honest there as well.

• 31:40Speaker 01

  Yeah, and reflection, introspection is really important.

• 31:43Auren Hoffman?

  Do you have any other interesting tells based on their resume or based on whatever? One of mine is, hey, if they have like a yahoo.com email address and they're applying for a tech company, it's almost always bad.

• 31:54Speaker 01

  There's a similar thing I've actually found to be good signal though, which is engineers who have Android phones. I've actually found to be very good. I don't know. It's not the same thing, but it's an interesting version of that. If people have more than two or three sentences on their about me section on their LinkedIn, it's a hard no for me. I will not talk to them.

• 32:12Auren Hoffman?

  Two or three sentences on the about me. Okay, interesting.

• 32:16Speaker 01

  It's just true. I've just never met someone who's great, who has more than two or three sentences about on their LinkedIn.

• 32:21Auren Hoffman?

  I like it. Okay. I gotta go check mine. I may have like seven sentences. So I gotta go check it out.

• 32:26Speaker 01

  There are exceptions.

• 32:29Speaker 01

  There's just not many.

• 32:30Auren Hoffman?

  I love it. That's really interesting. These pattern matches are really important because you need to make decisions quickly. And it's okay to occasionally just have false positives and false negatives, but you want to lower those ratios of both as fast as you can. If you have a false negative and you don't hire a great person, it is super annoying, but that will happen. You're going to miss out on some great people. And then of course, if you hire someone that isn't good, that's also annoying, but you can remedy that. But you can't make those mistakes at volume and you have limited time with people. You're looking at their resume in, I don't know, 30 seconds or something to make a decision whether or not you want to talk to them. You may have a quick test and other people are meeting them. It's just Stefan meets them for eight minutes.

• 33:12Auren Hoffman?

  It's very hard to make the assessments and you don't want to just outsource your thinking to what the admissions officer, when they were 17, decided.

• 33:21Speaker 01

  The other thing we do a lot is references and of course there's the references the candidate provides, but we also do a lot of blind references. Second degree references.

• 33:29Auren Hoffman?

  But you can only do them that at the very end because that's super time consuming.

• 33:32Speaker 01

  It's time consuming.

• 33:33Auren Hoffman?

  Right. You can't really do these upfront in the process. So when you've already made a decision to hire that person, you're doing the references of either to change your mind or to maybe help you. Cause what I love about the references, even if I've already a hundred percent decided to hire them, they can help me make that person more successful. Like, hey, if I'm the manager, how do I make this person really happy? And they might give you a little tip or here's something you should do. And the first few weeks of the person coming to your company is really important. And that might determine the entire trajectory of whether they're successful or not.

• 34:05Speaker 01

  Yes, that I think is definitely true. There's another version of it though, which we actually do earlier in the process. Because we hire predominantly from in-network, from referrals, almost exclusively actually from referrals. The way that we identify whom to interview or whom to court, because it is like a courtship, is from a lot of these back channel references to even identify the list.

• 34:25Auren Hoffman?

  Okay. Should you just send someone a text and say, hey, is this person amazing? Only text me back if they are.

• 34:31Speaker 01

  There's that version, but you do get into issues. You don't want to burn that person. That person might not have told their employer they're leaving, or they might not actually be leaving. They're just interested in talking. So what I've done is I will find someone who is willing to talk to me or someone I know at a company that I know should know this person. And I'll basically say, hey, I was looking at LinkedIn. Here are seven or eight people I was thinking about reaching out to. Any thoughts on any of them? And the person I want is one of them. And if they're not like, oh, yeah, this is one of the two best people at the company. Then we'll go, this is not worth talking to them.

• 35:01Auren Hoffman?

  Of course, if I was at that company, I would be like, if they're a great person, I'd be like, this person's horrible because I wouldn't want you to recruit them away. Right?

• 35:07Speaker 01

  Most people aren't that maniacal, though, or that diabolical. You and I are, but most people aren't.

• 35:13Auren Hoffman?

  Yeah, they're terrible. No, don't reach out. Don't reach out to Jade. She's terrible. Right? And she's like actually the superstar who's making the company. And then you also have a super small ops team as well. Even though you're managing licenses across 50 states, you're just trying to automate everything?

• 35:31Speaker 01

  Yes. So we tried to hire a head of ops for about a year and a half. And I didn't realize until pretty far into that of not liking and not finding good fit, that I didn't actually want a traditional head of ops. I wanted someone who's actually an AI engineer. So we hired an AI engineer as our head of ops. And it's been fantastic. And we have one and a half people on the ops team, a chapter, because we basically just AI everything. In order to sell Medicare plans to be a Medicare advisor, you have to be federally licensed. You have to be licensed in every state. You have to have contracts with every insurance carrier in every state. And then you have to do that for every single person. So this is a pretty big combinatorial problem. And each one of these requires different forms and annoying things to go through. So we've AI'd that entire thing.

• 36:15Speaker 01

  Most other companies in the Medicare space will have a team of maybe 50 or 60 people handling this. We have half a person.

• 36:23Auren Hoffman?

  Wow. Okay. That's insane. I assume you did that because it allows you to move much faster, make quicker decisions, et cetera.

• 36:31Speaker 01

  And higher quality, too. You have fewer mistakes. If it's all centralized, it's just easier to know where are the risks? Where are they not? And just culturally, we want to keep as small a team as possible as we scale really quickly. I don't want a lot of people running around who are just doing paperwork all day. And they don't want to do that either. It's hard to find good people who want to do that.

• 36:49Auren Hoffman?

  Okay, so a couple of personal questions. Let's go back to your heuristics. Okay, you have people who say they're passionate about something. You have the Yahoo email address. People have long LinkedIn bios. When you see somebody that has a hyphenated last name, what do you think about that person?

• 37:06Speaker 01

  I feel sorry for them because I know how painful it is. My last name didn't fit on the SAT Scantron.

• 37:11Auren Hoffman?

  Oh, right. Good point.

• 37:14Speaker 01

  Every second matters. So I spent more time filling in my name and had less time to answer the questions. But also, it just creates so much name conflict. Different systems, different health systems, different university systems, different state systems all have my name differently. It's very annoying, so I feel sorry for them.

• 37:30Auren Hoffman?

  Got it. Are you married?

• 37:31Speaker 01

  I am.

• 37:32Auren Hoffman?

  Okay. Does your spouse have a hyphenated last name?

• 37:35Speaker 01

  No, she has a three-letter last name. And I've joked that I'm just gonna take hers.

• 37:39Auren Hoffman?

  Yeah, that seems better. Because imagine if Blumenfeld Gantz married Bankman Freed. And you'd be like, Blumenfeld Gantz, Bankman Freed, or something like that. Which, by the way, sounds like an amazing law firm that I should hire. It could go on forever, right? Yeah.

• 37:54Speaker 01

  I think the reason I am pretty careful keeping the Blumenfelds is, so my mom's last name is Blumenfeld, my dad's last name is Gantz. And my sisters and I have Blumenfeld Gantz as our last name. My mom's dad was one of, I think, 11 siblings, and he was the only one to survive the Holocaust. So to my mom, it's very important, understandably, to have the name continue because most of the Blumenfelds were wiped out in Poland. That's the reason I very intentionally keep it and why I probably won't actually take my wife's name. But it would be a lot easier if I did.

• 38:22Auren Hoffman?

  All right, well, two more personal questions we ask all of our guests. What is a conspiracy theory that you believe?

• 38:27Speaker 01

  I didn't used to believe this, but more and more, I do think there's some element of consciousness being beyond our human bodies. There's too many things that happen in the world that seem like coincidences that just happen, that can't be coincidences. And too many, even scientific experiments, there's this one where you take a random number generator, which in theory has an even distribution of numbers. And then you put any thinking being, even a chicken, near it, and the distribution of numbers changes. It's wild. And there's this, experiments have been run at pretty high scale.

• 38:58Auren Hoffman?

  What do you mean, the distribution of numbers changing?

• 39:01Speaker 01

  The random number generator is no longer a random number generator. It skews to a number.

• 39:06Auren Hoffman?

  Oh, got it. So it's actually no longer very good?

• 39:09Speaker 01

  Yes, I know, these are very small margins, but at scale, you can actually see it. You put a human and you tell the human to think about really anything near the random number generator and it changes the output of the random number generator.

• 39:21Auren Hoffman?

  What? That's so crazy.

• 39:23Speaker 01

  It's crazy. There's all these experiments. It's hard to know how rigorous they are because we aren't there, but some of them have been done by tier one universities over the past 50 years. And the CIA used to run a lot of these studies where you have two people in different rooms. One of them is looking at a painting and the other one basically draws what that person is seeing. And at scale, this experiment has been run a lot. You actually get closer than you'd expect.

• 39:46Auren Hoffman?

  That's so cool.

• 39:48Speaker 01

  I don't know the reason, but the most common rationale there is there's some collective consciousness outside of our bodies that we all contribute to and pull from. There's something there that I think more and more is probably real.

• 39:59Auren Hoffman?

  Okay, I love that. That's so cool. All right, last question we ask all of our guests. What conventional wisdom or advice do you think is generally bad advice?

• 40:07Speaker 01

  I think planning your career is very bad. Plan your career far out is very bad advice. We have no idea what the world is going to look like. We have no idea what our personal preferences are going to be in the future. And I think people change a lot. Where I've always found that the best career advice, what I've followed, is just do the thing that's the best opportunity at the moment. Don't think about 20 years out. Don't try to create a master plan. I think that's how you get on these conveyor belts of going to a consulting firm or an investment bank because I'm going to go do that and then I'm going to go to private equity and then I'm going to go be at a venture fund. And it creates really bad outcomes.

• 40:43Auren Hoffman?

  Almost anytime someone is doing something that they think that will help them do the next thing is almost always bad. Yeah, I'm getting an MBA because the MBA will help me get this. Why? If you're going to get an MBA, like go there because it's going to be fun. Maybe you'll meet your spouse. Maybe you're going to have a good time. Those are good reasons, but to go there because you think it's going to like, I don't know, it just seems like an odd thing to go do.

• 41:04Speaker 01

  Totally agree. Yeah.

• 41:06Auren Hoffman?

  So really what you're saying is optimize more in the shorter term. If you're going to go do something for your career, do it because you find it inherently interesting. You're working with good people. Maybe it's a comp decision, maybe there's something else around that. Usually what people are doing is they're optimizing for the medium term, not the long term anyway. And so what you're really saying is don't optimize for the medium term.

• 41:27Speaker 01

  Yes, optimize for the short term because if you optimize for the right things in the short term, it should compound.

• 41:32Auren Hoffman?

  Okay, that makes a lot of sense. This has been amazing. Thank you, Koby Blumenfeld Gantz, for joining us on World of DAS. I follow you at Koby B. Gantz on X. I definitely encourage our listeners to engage you there. This has been a ton of fun.

• 41:49Speaker 01

  Thanks so much for having me, Arun.

• 41:53Auren Hoffman?

  Thanks for listening. If you haven't already done so, please subscribe to World of DAS wherever you consume your podcasts, YouTube, Apple, Spotify, and more. And please help us get discovered by leaving a review and check out worldofdas.com. That's worldofdas, D-A-A-S, dot com. And of course, connect with me on Twitter, at Auren. That's A-U-R-E-N. Would absolutely love to hear from you.