Garner Health Founder on Measuring Doctor Quality, The AI Landscape & What Improves Healthcare

追加: 2026-05-31

[00:00:00]This is Vital Signs, a podcast on cutting edge trends in health tech and the people shaping them. And today we have an episode, I don't know, 5 years in the making, uh, with with Nick Reaver, CEO and founder of Garner Health. Nick, uh, I've wanted to do this for so long. I feel like Garner is just this fascinating business on so many dimensions that maybe a lot of our listeners won't be super familiar with, and you're also just one of my favorite people to jam with on health tech. And so, I think this will be a lot of fun.

[00:00:22]We brought Nikil along to make >> I happen to be here, too.

[00:00:24]>> Yeah, Nil's here, too. He's great. Um, but really looking forward to getting to chat with you.

[00:00:28]>> Awesome. Can't wait to do it.

[00:00:30]>> A bunch of places we could start, but I feel like the origin story of Garner is really fascinating. And I know there's both a kind of like personal and intellectual or origin story. Maybe to start on the intellectual side. I know you'd spent some time postwater at Oscar Health and you'd kind of looked at a bunch of different things that were actually influential on healthcare costs and kind of came to a bunch of conclusions that ended up leading you to start Garner. I'd love for you to just talk about that kind of origin of the idea and journey.

[00:00:54]>> Also, more importantly, why is it called Garner?

[00:00:57]>> Yeah. Okay. We we can go there first.

[00:00:59]>> That should give you a good uh the types of questions that each party asks.

[00:01:03]>> Good cop, bad cop.

[00:01:04]>> Yeah. Well, I'll talk about the business, then we'll talk about why it's named once people understand it. Um so the so I mean, yeah, my personal journey, I had a bunch of um misdiagnoses and surgeries I ended up not needing and complications. And so that got me focused in on healthcare and I um left the investing world and um joined up pretty early on at Oscar was on the leadership team. And basically my goal when I was there is how could you create a business that would not just be a good actor in a broken system but actually fundamentally change how the whole system worked. And that was sort of the thesis that I was working on. And at the time I thought being a health plan was a great way to do that because you have all the data, you have all the information, you control all the money flows be a great place. And then what you start to realize that that's actually a terrible um thing because you have regulatory and a bunch of uh you know business constraints you don't quite understand. At least I didn't understand. And I think ultimately the the key was what I started to realize is um we were building one of the things I did was oversee our network development which is you build a insurance plan. you have a network of doctors and hospitals around the country. And what I started to see was when we were looking at the performance of our networks and our plans, what would actually change the needle for cost and quality. And by far the thing that mattered most, the intervention that mattered most is which individual doctor people were seeing. So we saw these like 4x difference in complication rates, hospitalization rates, adherence to clinical procedures by doctor. And some of those were at the biggest and brand name hospitals in the country. And so we basically then took this data and we're like, "Hey, hospital system partners, you're a great nonprofit institution with a great, you know, brand name for upholding high- quality care. Here's some amazing evidence that shows like a third of your doctors are actually lowering people's life expectancy um when they go see them." And they basically said, "Hey, listen, like love what you're doing, but those doctors basically are highest revenue doctors. They keep the lights on. so the other two/3s can do their job. And so, no, you cannot downrank those doctors. No, you cannot tell our your members that they're bad. No, you cannot remove them from the network. And if you do, we're going to drop out of your network and good luck selling a health plan in our state. And so that become basically the the puzzle is okay, if the key to this is there's such variability in basically the quality and cost of the product everyone's getting in terms of the doctor um and as a health plan you can't do anything about it. How do you create a new layer that could actually measure and actually drive difference differential behaviors um while not being a health plan? And basically Garner came out of of that work.

[00:03:45]>> So can we maybe double click to specifically like what Garner actually does today. So I know you guys work with employers and health plans to help identify doctors that are higher quality etc. And then there's some payment rail changes and network changes that happen as well. Well, but can you just like tell us a little bit about just like how it actually works? So, Garner, what we're really building ultimately is a new marketplace for how we think healthcare should work in the country >> and under the hood that really means we do three things. Um, we do doctor performance, we do demand and patient engagement, and we do provider improvement. And so on the on the data side, what we've done is we've aggregated this enormous data set of 320 million patients and we have mostly their full healthcare journeys um mostly in claims data although some other sets of data as well. So we and we use that to analyze every doctor in the country from primary care to subsp specialty cancer and who's actually delivering good health outcomes and following clinical guidelines and doing so at a reasonable cost in every insurance network. So that's the core asset. And then what we do is sort of two things on top is we one work with employers who are paying for all of this stuff, restructure their health plans and add a layer that sits on top of their United Etna Sigma plan that gives the right data and information and financial incentives to go see top performing doctors in the existing United Etna Sigma network. that changes meaningfully consumer demand and behaviors reduces costs a lot and then as we control we're now controlling one or two percent of US healthcare volume and growing very very quickly um we are now have a separate part of our business that then works with the providers to basically improve their quality and lower their cost and that then sort of completes a marketplace where you have you know you you get enough eyeballs shifting demand then you get to work with suppliers on improving their performance and you basically create a market around that.

[00:05:43]How many of the providers actually see their own data and they're like, I agree with this and we need to change it.

[00:05:48]Versus how many of them are they look at it and they're sort of like, we don't agree with this and we don't listen to your quality scores, blah blah blah, and like just move on basically.

[00:05:57]>> Yeah. Well, a couple things. So, one, um, that process has gone a lot better than I thought it would based on my, um, what I the little bit of experience that I had at Oscar doing this sort of stuff.

[00:06:09]And the um a couple things I think are are truth. The first is that what that by and large the um the way that providers have been evaluated has been totally unfair.

[00:06:20]>> Um and largely if you get into how this stuff has worked, it's all these algorithms that were built in the 1990s and it really does one number which is cost per patient.

[00:06:30]>> And you go around telling all the doctors, you have high cost per patient, you have low cost per patient, you have high cost per patient. Well, the people with the high cost per patient are obviously going to say my people were sicker and that's the and and turns out if you look at the data they're actually right about that that those methodologies are basically driven by that.

[00:06:46]>> And so what we did and we can get under the hood of this a little bit more is try to create a new methodology that was fundamentally not tied that could disambiguate provider sort of panel and who's walks into their door from their actual clinical performance. And when you do that, it turns out the providers are way more suscept uh, you know, open to that type of feedback. We now have some of the larger hospitals that actually use our data to change physician compensation and benchmarking.

[00:07:15]Um, is that to say that every doctor loves their score? Absolutely not. Um, and we as we get a little more vocal, um, you know, people were getting more, you know, questions and and all of that.

[00:07:26]But the nice thing is that we have a fundamental framework that in a detail assesses doctors on ways that are actually attributable to their performance versus randomness and that is a much healthier conversation.

[00:07:39]>> This quality problem is obviously incredibly deep and you have this like you know uh large data science team from all impressive backgrounds that are doing a ton of work on it. maybe help our listeners understand like an example of just like why this is such a complex or like the types of technical problems you have to solve to like actually be able to uh to really discern quality.

[00:07:56]>> Let's just take oh I don't know um knee pain. I I'm actually dealing with some knee pain now. I've been running for a little bit. I got to go see an orthopedist. Fine. Um so and I want to understand which orthopedists are the best for my knee. So let's start foundationally. Um, if I want to have a data set to analyze these providers, um, I'm going to have to compile data from many, many different sources. And we've got a couple hundred of different data sources, um, that we have to clean, validate, model, put together, stitch together, all of that. And it turns out even just doing the basic data modeling and all of and, um, validation and gating and checks to build this data set is actually very hard. It requires bespoke cloud infrastructure um, etc. So, there's that. um then I need to be able to figure out um something and it's somewhat easy somewhat easy to say okay cost per patient as I mentioned but but it turns out that um in order to really assess a provider you need to have a whole bunch of very specific things created so you need to understand I walk in um which prov turns out a provider can bill uh one of a five different level of codes and that initial office visit can cost either anywhere between $60 and $250. Well, which providers how do they do they do that fairly or not?

[00:09:12]That's just like one thing. Do they run do they give you an MRI when you need one, not when you don't need one? Um, what do they uh what do they diagnose you with? It turns out I have runner's knee and was that actually the right diagnosis and not and and all of these things. And I can keep going on to when do they do an MRI, where do they send the MRI, where do they do their surgeries, what's the complication rate of the surgery center versus the doctor, and what drugs do they prescribe, and all of these individual questions um all go into that one decision. And it turns out creating a sort of framework where you can in in some in a sort of um machine learning context engineer these features um at really wide scale where you need hundreds and hundreds of features um to analyze just something like orthopedics is actually really hard to govern maintain and then by the way run that performantly. Even a question as simple as um which doctors do physical therapy before surgery. Turns out running that in in an efficient computationally efficient way off across 100 billion healthcare records actually very hard technical challenge. So um you really have to build a whole new bespoke way of of collecting, analyzing and validating data in order to just answer one basic question called who should I see for my knee pain, >> right? because I mean and also people just present with so many different you know sub issues or different you know the the probably the surface area of the problem space is is so massive that uh it's probably hard to go you couldn't go like individual thing by individual thing because of just the sheer scope of the problem >> right and you need and what used to be the case is somebody would like write one rule called let's let's um figure out which doctors do physical therapy before surgery and that itself would be about a thousand lines of SQL code because you have to like you know look at all the who did physical therapy four or they got done that day and you have to have a longitudial record and whatever. And a lot of our sort of innovation has been the ability sort of our data engineers and data scientists basically build a platform on which clinicians can very efficiently do feature engineering on top. So clinicians can write in basically plain English or clinical English. Give me all the patients that look like this that have never had physical therapy before but also don't have a tumor in their knee but also do have this but also are between the ages of 18 and 65 and and then basically you know with that you know 20 lines of JSON you can basically very efficiently compute one feature and then that's one thing and then you times that by a thousand and now we're using AI to ongoing govern that etc. So it turns out it's a relatively tech uh challenging technical um problem and to to get out of the world of the like one data scientist write a thousand line of SQL code for one thing that doesn't really scale >> totally and then I guess on the flip side of it you're presenting this data to consumers right and I feel like there's always this debate in healthcare of like well how much you know information should we you know give consumers without overwhelming them and I imagine this is something you've thought a lot from the product experience of you know what's like the right level of detail to surface to folks about you know is it just a simple score. Is it the details behind that score?

[00:12:18]>> How how have you thought about that and how has that evolved over time?

[00:12:21]>> So I guess a few things. So right in our our platform we get almost about just about half of the people on the health plan to use our tool every single year which is about 10 times more than what you get otherwise um if you just like you know um give people I don't know their insurance directory or whatever a transparency tool. Um I guess we've learned a few things. The first is um that by far the most important thing is is money. Um and people care >> everything's incentives.

[00:12:51]>> People care as you know maybe not a shocking amount, but they care a lot about money and financial incentives. So if you tell someone you have a $4,000 deductible, by the way, you you know probably don't have $4,000 in the bank.

[00:13:03]You can go wherever you want. Or if you use our tools, your employer will pay you out of pocket. Well, it turns out most people will use tools. Um, so that's the most important thing. And then at Garner, the the challenge that we have is that we have an incredibly wide variety of customers. So we have Fortune 10 technology companies and we have like 50 life trucking companies >> which like most employer products have not served traditionally. Like they just that whole like lower part of the market I feel like has been, you know, underserved by a bunch of these.

[00:13:31]>> Exactly. And so like if you're going and we have like a bunch of um you know mostly Spanish- speakaking people who work in construction and you want to get them to make a different healthcare decision, it's it's actually an interesting and somewhat different product um experience than the people who like I'm a data scientist at a Fortune 10 technology company and I care about data. Um and so what we found is a couple things. one um you have so setting up a basic financial platform really helps and then in terms of the information you basically want to give people the ability to click in as much as they can and have um sort of a digital self-service experience and a human experience and what we found is that basically some people are like this doctor is good and that's good enough for a lot of people we think this doctor has better than top 20% health outcomes great I'm good and some people want to know the hospitalization rate of the doctor because their spouse got diagnosed with cancer and they want to know the mortality rates or whatever.

[00:14:34]And so what we found is that basically you have to structure your data in such a way that there are varying levels of aggregation that can be clicked into and then different users have the ability to access clicks down as much as they want to. Um, so that that's really how we've architected it and we've learned a little bit the hard way that if you don't build the systems the right way, you won't be able to do those aggregations and then you'll you'll lose some or the other users along the way.

[00:15:00]You've talked a little bit about like the variance in employer types that you guys serve and then obviously even within those employer types there's variance in employees. Um, you know, I think the debate in healthcare has always just been like what people consider quality is going to be very different from person to person, from entity to entity, etc. Um, do you guys let employers also pick their own facets of the quality score also? Is it sort of one score that's used across everyone?

[00:15:27]And then how do you think about, you know, what, you know, people think about highquality care from maybe the employer side versus the versus the patient side versus the doctor side, etc. Because I can imagine a scenario, you know, where patients like, oh, actually, I want the doc that does imaging every single time to like make sure that everything's okay, right? even if you don't understand like context and all that kind of stuff. Yeah, >> I think um as we've gone through it, there's >> our view is something like there's a few different versions of how you might define quality >> like do I care about life expectancy or like healthy days or you know things like that like quality adjusted lifespan things of that versus >> Jacob and I are both huge longevity guys so this is very important on our Peter >> yeah we'll go deep um what's interesting is that mostly most of the debate is not so much in in like sort of what is the measure of quality, but rather it's that people have crummy methodologies that don't actually correlate with life expectancy or quality adjusted life expectancy. Um, and so our goal has been to take as much as we can all of the guesswork out of those methodologies and as much as we can u make it transparent.

[00:16:39]So let me take your example of oh I don't know uh you said MRI some you know I don't know I go to a cardiologist do I want the cardiologist that gives me all the tests you know um it turns out you don't want that and it's relatively an objective fact that you don't want a doctor that's going to go like run you through the mill of tests >> and we can maybe I guess we can debate that but there's a bunch of academic literature that shows that if you get an unnecessary echo cardiogram your life expectancy goes down and not up um and and the Reason for that is because there's a false positive and if you get an unnecessary echo cardiogram or stress test you're much more likely to get invasive angography which is just a more invasive type of diagnostic test and then if and that that has a false and by the way the echo cardiograms about a 25% false positive rate the stress test the um angography has about a I think a 10 to 15% false positive rate and so you can like one or two% of the time go from false positive to false positive then you're going to go have yourself a surgery Y um and then you're going to have you know about 10% of that depending on what you get is going to have a is going to have a complication.

[00:17:46]So you know something like 20 basis points at a time you're going to have a serious complication just due to false positives and that is not I mean we can debate that but that's actually just math as shown in the literature. Um it's sort of like people who like I don't know baseball or basketball like you can debate taking three-pointers or two-pointers but it's just math at the end of the day. And so that's how we do it is we just at every measure every decision in terms of we choose quality adjusted life days and its impact and we think the literature is actually relatively unambiguous on those topics.

[00:18:16]But the problem is no one has taken the time to meticulously add up the the positive of the doctor doing the test at the times that it's needed which is needs to happen against the negative of overdoing the tests. And then by the way adding that to all the other different activities in um in a patient journey and coming up with a number. And so if you do that though you get something that you know is relatively unambiguous.

[00:18:40]Um we can debate one number is out of 10% or 15%. But by and large we think that there's we measure quality adjusted life days for every every um patient journey and we measure it in this very meticulous manner and it takes a lot of the ambiguity out of the process of how do you define quality. Then you mentioned obviously you know you have a big part of the business that then uh works with health systems right and helps them think through this and helps them you know uh oversee their you know uh provider footprint. What have you kind of learned around like what are the most interesting systems doing with this data or how are they thinking about you know the most forwardleaning ones what they how they might change the way they operate. Yeah, I mean it's a it's a incredibly fascinating time in US healthcare because you pro everyone's heard that costs are rising incredibly.

[00:19:26]>> I've heard I've heard rumors charts. Um it's happening. It goes up and to the left um up into the right rather. Um I wish um and so what's happening is basically the government and employers are getting squeezed and then they're pushing hard on providers and that means providers are getting squeezed which means the core economics of how do you run a hospital system is not the same as it was. Um that means that basically that there's a good chunk of systems and we're the sort of idea of Garner is to accelerate this is a lot of systems starting to realize wait employers are now designing their plans to reward employees who are seeing highquality doctors who aren't running the mill on surgeries and who have low complication rates. It used to be um that providers would make double if there was unnecessary surgery and triple if there was an unnecessary complication. More and more systems are starting to realize that because employers and the government are changing their plan design that the only way they're going to compete 3 years, 5 years, 7 years down is to compete based on quality. And and so we were starting to see more, you know, I would say 10 20 30% of the system starting to say like I have to proactively position myself as the highquality, lowcost alternative and that's how I'm going to compete and that didn't exist five, seven years ago in in our healthare system is nobody was thinking about that.

[00:20:53]>> Yeah. I mean I thought valuebased care was supposed to bring us here. So like what you know I guess what you know what didn't work maybe about this first wave of uh of options that now is starting to work uh so well I guess on you know influencing commercial patients. Yeah, I mean I think people I think um it has sort of been oddly gospel that value based care will get us to some lower cost uh system and the idea of for those who don't know is is that basically currently we mostly pay doctors per procedure. Um and the idea is well let's just and that encourages them to do more procedures right um so let's just pay them one flat fee per patient and then of course they'll not want to the basic challenge with it that is it's a repeat game um and if I do that as a and I literally had these conversations and I had large hospital systems say we love what you're doing we don't want to do you know value based care because like ultimately even if we're in it or out of it we're going to get rewarded by how many patients walk in our door in 3 years 5 years 10 years it's market share and if if we do value based care this year and we do a great job, we're our benchmarks are just going to get better and next year we're going to make less money. So it doesn't make any so ultimately we want as big a pie of healthcare doctors at dollars out there and we'll just take as much of a pie slices of that as we can. So the idea of a in a repeat game where you can re redo the price every year um value based care just doesn't work and the only times that it does work and is when providers have to compete for patients. So if you where is where is actually valuebased care working it's in like Medicare advantage. Why does it work in Medicare Advantage not because Medicare is special in some way and commercial it's not? It's because patients are individually elected decision about which providers they go to and if there's basically a better plan a lower cost plan they'll opt into that more or less and and it'll drive patient volumes and then those providers get rewarded with more volume. And so I think the whole thesis of how you will fix health care needs to be basically how do you change patient volumes.

[00:22:51]That's by the way like you know doing all you can eat at I don't know Denny's doesn't improve the quality of the food at Denny's. You would never think about that. You know you the basic point is you you know restaurants only compete by getting p customers in the door and healthcare has to work the same way and that's sort of the only way that we'll we'll fix the thing and that's basically why we've been focused on that and not value based care. I think in the player employer space over the years, you've seen different versions of like centers of excellence, bundle payments, like hey, we'll fly your people out to Mayo Clinic so that they can get it all done there instead of a local um whatever orthopedist, etc. >> Um and you know, there's different versions. Some of it looks very similar to value based care. Some of it is just hey, we'll just cover your costs and go there. Do you guys um think about having maybe more complex contracting when it comes to the provider side of things or is your opinion and view on this like hey we're just going to focus on the quality and pricing stuff and it's going to be straight free for service normally but we are just going to drive volume to specific providers and then that's kind of like our our belief on how we we make things the economics better here. Our goal is, again, it's sort of like we've learned with Google, Amazon, Door Dash, all these sorts of businesses. If you control patient eyeballs and where people buy stuff, um the suppliers have to get with the program. And so our on the demand side, we're mostly focused on changing changing how people make decisions. You know, I wonder like if I'm if I'm, you know, a patient and I'm shopping through Garner and I'm looking at uh whatever Mayo Clinic, Cleveland Clinic, etc., for like a name brand I've heard of in some capacity, and then I see, hey, I can go to this cheaper option that is close by, maybe I've never heard of it, etc. Like, how many people um actually do think about like, hey, uh I'm going to go to the cheaper option here. It's lower price for me versus a name brand that I've actually heard of >> uh and is local and etc., right? Which might be more expensive, >> right? Well, so first one thing I should note is that on our platform if PE regardless of how low cost a provider is um they they do need to meet our quality thresholds of a standard of care. So if they're not delivering better than average care, they we will not recommend them to a member. The nice thing about that though is that in in healthcare, high quality leads to low cost because you're healthier, you lower your cost and therefore you need less complications, lower total cost and that's how we get around the sort of like crazy Yeti and whatever. And by the way, if you start doing that, you have eye complications and it turns out crazy is very expensive, you know.

[00:25:27]>> Um, in terms of the brand name, I mean, the reality is like you try to analyze this and there is a small but but not very a small positive but not very large correlation between brand name and outcomes. And most people don't realize the re there's a bunch of reasons for this. The biggest one is that all of the big brand name systems bought all of the other systems that are nearby. So like name your top 10 hospitals. They're not even they bought a whole bunch of others and like I won't throw any specific systems under their bus, but they all have bought many more hospitals than they had. They have this sort of relatively small main campus which is the academic medical center and then they have 12 other hospitals that they slap their logo on. Yeah.

[00:26:09]>> Um and by the way, those other hospitals aren't even on the same instance of Epic.

[00:26:13]>> They have no similar standards. All they did was buy those hospitals, raise the price. It's like when they're like I'm blank affiliated and you're like what is what does that even mean? Are you owned by them?

[00:26:22]>> And and by the way, the main campuses aren't even that much better. Um they're not even that much better if you look at basic routine care. What they are better at is for very rare cancers, very rare things like go find your coronary academic medical center. But if you want a knee replacement, like it's not actually many any meaningful amount better at the main campus of an academic medical center versus your local community hospital. There's a lot of um types of care that are obviously not delivered by one physician, right?

[00:26:47]There's like a team behind the physician or you have access to the facility which has access to certain things. So you have the physician individual physician kind of quality scoring system. Then there is the like infrastructure that enables that physician to do good work.

[00:27:04]>> When you guys are thinking about scoring like how do those two things sort of interplay with each other cuz you might have a great doc that is has their own practice etc. but maybe doesn't have the like team around them that helps them do excellent stuff maybe in certain procedures or whatever, >> right? And so there's ultimately when we recommend doctors, we think about them in light of all of the other um downstream resources that they utilize or the physicians with whom they work.

[00:27:33]And that may seem sort of unfair, like the best doctor in the world could be working at a terrible hospital and we won't recommend them if they're if the net of all those things is bad health outcomes. Um, but we think that that's what we owe it to our members to do, which is it's not doesn't matter if that doctor's great. It's the net of what physicians they they partner with, their PAs and NPs and people that work in their office. Um, by the way, even the office front desk staff, if they give a terrible experience, that matters, too.

[00:28:03]Not to mention the hospital where they do the surgery, the anesthesiology group that staffs that, the pathology, radiology, the MRI to whom they refer, and on and on. And so, it turns out the most important of all those things is almost always the lead physician because they're deciding everything else, but they are far from the only thing that matters. And basically our systems what they do is analyze every one of those in things independently. How good is this surgery center? Um what's the cost of this MRI center and on and on what's the quality of it? And then it's basically those people who maybe I'm dating myself but it's like the price is right Plinkco board where you put in the thing and you you know it like B bounces around and you inherit all this decision tree of all these other things and we can basically watch from a statistical perspective the Plinko board unfold and we know this doctor will use these physicians and route to that MRI center and this surgery center and it adds its way up in into one overall score.

[00:29:03]>> Yeah. So you obviously started the company with these like you know really innovative quality algorithms and I think then over time have built like uh a really strong consumer app and front end and have been really focused from a technology perspective on just bringing a really seamless experience to you know end customers because as we've been talking about in this conversation it's all about you know how do you get end customers to trust Garner use that to figure out their care. Maybe talk a little bit about like it's such a large product service area. What are some of the tech challenges involved in in building that service area and what gets you excited about the you know the frontiers?

[00:29:35]>> When we started we sort of knew that there were companies that had not maybe in the best way analyzed doctors who you know insurance companies have been doing this for years.

[00:29:43]>> I remember before we met I'd heard about what you were doing and I was like okay well it's either like you know I've seen cast light I've seen accolade uh you know I guess I'll I'll meet this guy but I don't know he seems smart I don't know why he's going down the space and then uh within 30 minutes I understood. Thank thank good feel damn >> I appreciate y'all's support. Um the uh well I mean I I think the the reality is a few things that there have been sort of data companies um in healthcare.

[00:30:10]There had been sort of patient navigation apps to search and find care.

[00:30:15]Um, and there had been financial administrators and basically and not great ones of any one of those three unfortunately, but we basically decided that the minimum viable product was combining all of them. And so you had to get the data right, you had to have an amazing user experience um and you had to get the financial incentives right.

[00:30:35]And so what that resulted when we launched, you know, we basically um had to cover a massive product footprint with a very very small team and it was very challenging. Um and but we we felt like we had to do all of them otherwise we wouldn't get the results that we did and our early client results ended up being very very good. Um and so you know ultimately it's been a little bit of undoing the duct tape and bubble gum that we created over the last six years and really making it work. I mean if you just take um and and I think also the emergence of AI gives us um a whole new leg to go in terms of where we go. So if you just take a question like okay let's imagine um we know who the right doctors are and we've talked a lot about the challenges there and there's a whole bunch of things about getting I AI to read medical literature and and automate some of the research work we're doing but let's imagine we already know that.

[00:31:27]Well, then turns out appointment booking and appointment availability and even the phone numbers are an incredibly hard challenge. The insurance directories are 61% inaccurate, meaning mostly wrong.

[00:31:39]Um, so how do you figure that out? And um and it turns out having an AI that can actually make phone calls and book appointments for you is a a big piece of the challenge. an AI that can surf the web and actually click around on through every doctor's uh website and figure out their phone number and click in and see when they have appointments availability is a big part of it. Um and then there's a whole machine learning to to grab other data sets and put that all together etc. So that's like just one thing well one little decision that we then had to build um and are continuing to build and then uh then you have all the user front end and you know how do we become the most and I think we have a place to play there the most trusted place for people search understand everything from hey my toe hurts what should I do to I just got diagnosed with this rare form of cancer so all of the the front end and user tools there are very complicated um and then making the payment rails work with every insurance company in the country um with increasingly having everything from HSA compatibility to HRA which is a different way of administering it to having a card to swipe and on and on. So there's just so much to do to make a really really simple thing work and I think that's why the early iterations of this stuff didn't work is they chose one of those three categories and um and they didn't really even necessarily deliver on that. But you need all of these things to work in concert if you're actually going to change consumer behavior.

[00:33:05]>> Yeah. Well, you brought up AI, not me, but because we're, you know, a venture sponsored podcast, we are we are contractually obligated to, uh, to discuss a bit. Uh, I know you've been at the forefront of thinking a lot about how, you know, we're all going to use these tools going forward, particularly, you know, uh, from the front door and then throughout our kind of care journey experiences. Maybe I'll start there.

[00:33:22]Like, what do you think, uh, the average person's interaction with these AI tools over the next, you know, five years in their healthcare journey ends up looking like?

[00:33:29]>> Yeah. Um well I mean it's it's fascinating. We we passed in like two months ago we passed the point where um people are using AI more for primary care than humans for the primary care.

[00:33:40]So like that just happened. Um and that's only going to get more extreme.

[00:33:44]Um a couple things I think are interesting about it. Number one is if you look at healthcare dollars um you know healthcare interactions you can think about interacting with AI on that but all of the dollars are still nowhere until we get the robots not really touched by AI. So about 8% of all of health care cost is physician office bills like interactions that happen in an office where you could imagine an AI doing that and about only half of those at most you could actually not have a human because it's like an orthopedic visit where I got to go like touch someone's knee. Okay, AI can't do that quite yet. Um, so it's really only three maybe 4% of healthcare dollars that we think are automatable via AI interactions, but it's a huge percentage of the actual time that people experience. And so that's sort of where we're going is to become the sort of the trusted place where people go for all of those things that can be handled um by a human talking um and and the things that we have over oh I don't know other you know um the foundation model companies is we have all of the data trained on which doctors are high quality and we have the money um and we have appointment availability um and so our sort of vision is that people will need a healthcare specific AI that helps them understand their issue, navigate the health care system, what's my diagnosis, what should my treatment pattern be, and that more and more the off physician office visit will die down, but you will still have the vast majority of healthcare expenditures being procedural procedures and drugs and those things will take at least 10 more years, I think, for the robots to come get. Um, and in the in the meantime, sort of that's what I think where the world will go is you'll have one AI that will just help you understand your labs, diagnose you, etc. And then we'll help you route and find the right procedural list or the right drug to take when you need it.

[00:35:45]>> And there's been some version of that, you know, every few years someone's like, I'm going to, you know, create Yelp for doctors and it happens again.

[00:35:52]And I feel like it never quite sticks, right? that I think that business model is very tough to go in the direct to consumer route but it does seem like it's changing a little bit now with people now going to claude going to chat GPT asking hey which doctor should I see how many people one do you think that there's ever a regular just consumerf facing version of your business that doesn't go through the employer route that goes direct to consumer um and then two do you think that people will know when to go to Garner to ask the questions that you're suggesting they ask versus just going to the regular old uh LLM that they use on a regular basis and just be like, "Hey, just tell me the best doc to go to and then they'll know they will know, oh, this is not trained in the medical data and all this kind of stuff and actually like know to go to you guys." It's it'll be fascinating to see how this all plays out. I mean, ultimately, if you've seen, you know, the Super Bowl ads, most people don't want um an AI that gets paid with ads to um recommend them, particularly healthcare stuff. Um and and the even further than that, most people want something that I think is integ $4,000 deductible if you don't, you know, go to the if you go to these set of doctors and it's free if you go to these set of doctors, you're going to need that integrated in. And so ultimately there's, you know, if you take, I don't know, where do I get my knee replacement in New York City, that's a, you know, the the difference between the top cortile and the bottom cortile, it's $30,000, um, in terms of the total cost, including complications, etc. So, um, that, you know, that's a lot of money that whoever somebody's paying for that. It's either, you know, it's your employer or it's the government that whoever that plan sponsor is is going to want to create a design where you get some of this money back if you do the right thing and you have to pay more if you don't. And that is going to be a really challenging thing to integrate into a foundation model. um and um along with I think all of the detailed data on complication rates. That's just not data that's available on the public internet. Um and so my sense is that what is likely to win over time is something that is much more tightly tied to the health plan or is paid for by the health plan, the employer, the health plan, etc. Um because again I think people care about money and they care about the actual data on performance and so I think that's where it will go. Um you know ult and ultimately we've we're looking at and thinking about releasing some of our data to you know more of the foundation models. Um and we may do that as well.

[00:38:36]Um, but I still think ultimately people are going to want to go to something that is aligns with their plan design as distinct from something that may or may not be selling them ads on the back end.

[00:38:46]I think to corroborate your point for what it's worth, I had to go find a new doc recently and I did use a bunch of the LLMs to do it and then you get to this point where you're like, wait, is this a network or not a network? I have no idea. And I had to I ended up having to go anyway to, you know, I can't think included or something like I'm like, which one is it? the the the the models that can solve all coding tests uh known to man and math Olympiad problems in network or out of network. It's terrible.

[00:39:14]>> It's not in the training real.

[00:39:14]>> Yeah. It turns out you need all the all the context.

[00:39:17]>> Exactly.

[00:39:18]>> Um I guess like you know even um as you think about your own AI efforts which you guys have done so much on. How have you thought about what makes sense for you guys to build versus you know hey look the models are going to keep getting better or people are going to build these other things like you know we we don't need to be especially good at at X or Y. Yeah, I think for the most part um my I mean the the general principle we've seen is that general models are going to win. I think um at least as it pertains to things like diagnosing care and figuring out um yeah, what is the accurate reading lab results, figuring out what the right treatment pathway is, what is the diagnosis. Um, I think that the jury's I think in on that and I think the people that were spending hundreds of millions of dollars trying to fine-tune their own model around like you know trying to more accurately diagnose the disease I think that stuff is available more or less in the public internet and um and anything available in the public internet can be trained on and it will work and anyway so we we think we will basically get all that stuff for free um and we are setting ourselves up for that in other words that things like a real deep clinical expertise um and the ability to diagnose members uh we think and thus a lot of the member-f facing things are going to be free uh as I mentioned the things that won't be is all the tools you give them your insurance directory your plan design um all the stuff not on the internet around complication rates um and then we're doing a bunch of stuff that's a little more um bespoke on internal tooling around how do we create our own AI clinical researcher that is a little bit better at reading clinical research and writing code for us. And there's some stuff in the intersection between basically like writing Python code and reading complicated academic literature and translating one to the other that we've actually found the need to fine-tune some models a little bit on. Um again because that stuff's not in the public internet. It's bespoke and hard. So most of our real machine learning work goes on the back end. Um and then our our front end things were were much more leveraging tool use and and and sort of riding the wave as the models get better. You can imagine over time like you know getting you automate more and more of this you get more and more tailored recommendations and I assume you know given these unique data sources you have too that like you have this like almost constantly updating flywheel of of both literature but also what you're observing from outcomes and and and real world studies uh that then make you know these recommendations you know even better over time.

[00:41:53]>> Yeah. Exactly. Well and and by the way as we have more deep relationships with providers that aids it as well. So we now the cool stuff is we now are able to from for many of our provider partners get all of the data the EMR data the lab data the claims data our data on patient engagement and have one unified like really true longitudinal health record um for those members and that is an unbelievably valuable training data set and also because we now have AI you can now have you can now get over the all the EMR data is unstructured problem and so you can basically basically have the AI create a structured outcome of your EMR record, compare it to your claims data record and our internal sort of app usage and other other data of that nature and then you have something that is really powerful to assess physician quality even even better than before.

[00:42:45]>> Yeah, I'm always struck by it feels like the most badass healthcare businesses are, you know, it's like you have to understand healthcare at like the 10D layer that you do, but then also have this like rely toward technology and like actually implementing this stuff versus, you know, classic putting it on slides. And I feel like you guys uh combine that quite nicely. We're working on still a lot to do.

[00:43:01]>> All right, let me throw out some random data sources and you tell me how good you think they'd be at figuring out the quality of positions.

[00:43:07]>> Okay, I've never played this game before, but that sounds great.

[00:43:10]>> This is a game we're making up on the spot.

[00:43:11]>> I'm excited. What random data sources are we talking here?

[00:43:14]>> Well, okay, let me put it this way.

[00:43:15]Like, for example, I think referrals are sort of this interesting dark matter uh uh data set, right? where it's like um docs who tend to refer a lot of patients of a specific type to a certain doc.

[00:43:28]>> That doc is probably halfway decent at that thing. Or maybe it's just that they went to med school with that doc and they just have like a relationship with that doctor and they're they don't actually know if it's that good or not.

[00:43:39]Or for example, if you were to like uh poll anesthesiologists about how good a surgeon is actually like you would probably get an interesting map of how good different surgeons are at different surgeries. So there's obviously like the raw claims data and all this these sources that are probably pretty obvious, but I'm curious like are there esoteric and random data sources that you think actually might be good predictors of quality that aren't used as frequently or >> are you proposing that like garner AI interviews like every doctor and it's like who are your exact just secret shop and just like you know go to the anesthesiologist loun and be like yo you heard about Dr. Jacob, he sucks, right?

[00:44:21]You know, like that should be a data source included.

[00:44:23]>> Well, I guess yeah, so we've thought about this topic. It turns out, not quite in those terms. Um, and so a couple of things. So, one, if you look at things like so, we can measure referrals and who refers to whom. Um, interestingly, for what it's worth, is that referral patterns have basically broken down over the last decade and that 75 or 80% of people self-reer now.

[00:44:45]Um, >> like back to their own hospital.

[00:44:46]>> Yeah. you were out on asking chat GPT for a doctor, right? And so most people don't have a longitudinal primary care doctor. Um, and even when they do and they get a referral, they still go online and like, I don't know, let me go look at the Yelp reviews for this doctor or Google reviews or whatever. So the the nature, so a lot of the traditional sources don't even correlate. They don't there's not even stability on referral patterns. And when they do, they correlate much more with golfing relationships than they do with other things.

[00:45:14]>> But um, there's that. um you know the one that the one that's also out there that people say which I think could be interesting is you if you did a survey of the nurses that worked in the hospital >> would that correlate and whatever the interesting thing um is sure but also wouldn't I just if I could know the complication rates wouldn't I just look at that like I if if in times when those things disagreed like do the nurses and you had a really good nurse survey or anesthesiologist survey wouldn't you just kind of go with the complication rates and like the actual data of the health.

[00:45:49]>> It's like the stats revolution you're talking about in sports. It's like, you know, for a while you had the scouts that was like, oh, I like I like the look of that.

[00:45:54]>> One more money ball reference in healthcare. I'm going to jump. But anyway, but I think that's the point is like some of the I don't know some some of that stuff would and it turns out like yeah like there's a little bit of a relationship between years of experience. there's like a U curve where um you know the you're earlier on you tend to have worse outcomes and then when you're later on you also tend to have worse outcomes. Um we haven't used much of that stuff at all because if you can just look at the actual outcomes it's just better. And so um there's a bunch of coralates that you can use but we end up just using the real stuff.

[00:46:28]>> Is there ever a world in which you guys are setting the foundation for a physician to leave a hospital and like become independent? Because you talk to physicians, like one of the things they're always most concerned about about going independent is are patients going to come my way, especially if I don't have a hospital or people aren't referring to me. Uh my contracted rates are going to be worse because I don't have size, all this kind of stuff. You can sort of solve like a few of the bigger issues that physicians are worried about. And that is ultimately like what does bring costs lower for employers and payers is people with more competition in the provider.

[00:47:03]>> Totally. Yeah. And that's what we're trying to create. I mean it might not be crazy Eddy's but we're trying to create the sort of like you know nice high quality version of that is people splitting out and the issue as you mentioned today is you cannot even if you invented um oh I don't know a surgery center that had way lower complication rates um and cost a lot less you can't fill it um and by the way the insurance companies won't give you any good may not even bring you a network at any reasonable um reimbursement rate so you can't start a business today where if you have a way better product at a way lower price. You cannot deliver that in healthcare. Um and so um the nice thing though is is with with what we're able to do is now in a lot of different markets, we're you know 3, four, 5% of the patient volume.

[00:47:49]And so if somebody wants to start, you know, get at us if you want to start a surgery center or whatever, get at us.

[00:47:55]Um give us a call and um we could probably fill it up for you. And we could fill it up, by the way, with commercial employer patients that pay three times more than Medicare and Medicaid. And so that's definitely the future that we want to build. Um I don't think we're going to do the like uh Amazon Essentials and like, you know, build a bunch of surgery centers ourself because of the a bunch of reasons, you know, but um >> never say never. Jacob's right here. He can bankroll it or think, right, personally, if nothing else, >> depends how big this podcast is.

[00:48:21]>> Yeah. Yeah. Exactly. Um, so the but I mean we've thought about that a lot and we are starting to we really want people the to create business models where people can spin out launch businesses.

[00:48:34]We'll fill them up as long as they do a great job with quality and do so at a low cost. We'll fill up your panel all day long. Um, and you don't even need to really pay for ad spend. Um and so so anyway that's the basic vision of the business and whether we like start to create franchises and partnerships on that we'll figure that out a little bit but but you know as we've been doubling or tripling every year you know we're not that far away from having that in most every market in the country. You're obviously this you know deep student of of business models and healthcare and you're seeing all this AI stuff. I'm wondering you we were alluding to it earlier that basically you know hey it turns out you may end up like uh you know the valuebased care thing. It's like you may end up uh trying to lower cost and then you just get renegotiated every year on it. And I feel like I can see a lot of analogies to a bunch of AI businesses that are popping up, you know, promising to to do the same thing.

[00:49:20]As you look at the crop of just AI healthcare companies out there, what business models do you think are actually going to work or like what are the stuff is actually going to have legs?

[00:49:28]>> I mean, I think ultimately you need to use the AI. I think a lot of the businesses that have gotten good traction in AI, you have like two to three years before everybody commoditizes you. And it that's a little slower than maybe in other places because you can have a first mover advantage and there's real um it it takes a little while for everybody to hospital systems to adjust or employers to adjust. Um but ultimately you have to get something that creates a tool um to feed into your AI that has some sort of um connectivity to humans or unique data asset. Um and we sort of view that as you know on our side um we think a lot of >> better have both I guess >> better to have both is what we're working on. Um but the that ultimately get like you know you can get going fast and grow a lot of revenue as we're seeing fast with a cooler snappier you know UX wrapped around an AI model and whether that's revenue cycle management whether that's patient care navigation you know a bunch of others that are are are sort of going patient navigation and engagement phone calls um appointment booking that's great all that stuff will be a commodity by you know 2030 for sure probably faster than that and so you basically have two to three years with your hair on fire to say how do I then leverage that to really am I going to be able to get a unique data set or am I going to be so deeply ingrained in the human workflows of a hospital system of a payer that like it's really hard to rip me out. Um and I think that I mean that's an overly broad thing but um that's sort of what we're seeing and I think there's probably a few of the businesses that that feel like they're going to get deeply ingrained enough that this will make sense. Um obviously we we as we have unique data and we have ties via workflows and financial workflows and we feel like we have a good good play there. Um but you know probably not all and there's going to be a lot of tide that goes out and and folks figuring out what their actual business is.

[00:51:35]>> Yeah. No, I love that. Um we always like to end our interview with a quick fire round where we throw our favorite like overly broad questions that we stuff into the end of the interview. Um, and so maybe to start like, you know, if you had a magic wand, you could make one policy change to improve healthcare in the US, what would you do?

[00:51:50]>> Infinite deductible plans. Um, right now there's a limit.

[00:51:54]>> Isn't that just cash?

[00:51:56]>> Yeah, I think I do think the um like we're starting to run into this. I mean, there's a bunch I like data sharing. I mean, I like there's a bunch of restrictions in how do you state level uh policy stuff that I I think is painful painful and and why are we regulated at the state level. Um, but I think, you know, I think, um, a fun one is why we're starting to run into max of of deductibles and, um, we think that that's kind of silly. Um, why don't we allow people to have a $50,000 deductible and but, um, but then, you know, give incentives to people to bring that cost all the way down. And so, you're sort of there's this cap on the consumerization of healthcare, which is a little silly. It'd be relatively simple to fix. There's a lot of other massive things, but that's one that could actually reasonably get changed.

[00:52:42]Uh increase those maxes a lot and um I think it would actually really aid consumerization and bring a lot of good around.

[00:52:48]>> Interesting.

[00:52:49]>> Yeah.

[00:52:49]>> Um you mentioned the U curve for age.

[00:52:52]Are there other interesting things you've seen in your data set that kind of correlate around quality? Like either geographic splits or like you obviously probably sliced and dice this data a bunch of different ways. Any like fun anecdotes you've learned about about this? Um, oh, I don't know. Um, the the biggest negative correlate to quality that we've found is being the chair of a department. Um, it works. So, don't go to the chair of the department.

[00:53:22]>> They're out of the game.

[00:53:23]>> And I'm sorry if anybody's watching this that is a chair.

[00:53:26]>> You as a chair of the department are good. It's just the other ones.

[00:53:28]>> Well, you know, how do you get to be the chair of the department? Not all the way time, but usually it's because you have the highest revenue. What does that mean? It means you had pretty high complication rates and high, you know, like and you had pretty high um, you know, moving people right to surgery.

[00:53:42]And then so you take these doctors that are biased towards being worse than average. And then you basically take them out of the um clinic 4 days a week and you cut their volume by 80% and they get worse because they didn't do the surgery. They do the surgery once every like 3 months.

[00:53:57]>> Yeah. Yeah.

[00:53:58]>> So those people actually turns out are the worst.

[00:54:00]>> That's funny. I like that. Not all the time, not always, but but generally not great.

[00:54:04]>> I love that one. Great.

[00:54:06]>> I mean, one thing I think is fascinating about you're obviously so deep in healthcare now, but you spent, you know, most of the early parts of your career doing all sorts of big roles at Bridgewwater. I'm curious like what what did you take away from that experience to like healthcare?

[00:54:17]>> I guess a couple things. Um, yeah, there's a on the culture side, there's a lot I love about Bridgewater. There's a lot of we do not do the same way. And so, we've made some real changes on that. And I think giving people real autonomy um sort of balancing um how do you make sure you have all the right people in the right roles with in a high performance culture with care, empathy and teamwork. I think some of those things and then I think generally um it has become if you're really going to solve a societywide problem, it's very useful to really understand the economics of how the system works. And I think a lot of businesses fail largely or maybe they succeed but they end up being aligned less with the forces of good than they thought. Um and I think it's largely because people don't really deeply understand the sort of over the long term really how the economics work.

[00:55:10]And so that I think has been advantageous and we try to have not just on our corporate strategy but actually every product that we launch really understand the economics of this feature of this component of what we do. Let's take your average employer then just you're a Fortune 500.

[00:55:25]>> You are dropped into the seat of of the CFO of a Fortune 500 company. They're like, "We got to handle our healthcare spend."

[00:55:31]>> What are like the first things that you're looking at, changing, figuring out, etc. Since I feel like today a lot of Fortune 500 companies like obviously are thinking about this a lot more as their premiums are spiking.

[00:55:43]>> I mean, it's fascinating. Um the I I believe it's actually really really really simple and people just don't want to make any sort of change. So if it used to be at business travel in the 70s and ' 80s, you used to basically just cover whatever you do, you know, and you resulted in these ridiculous expenses and you know dinners at the fancy steakous and whatever and people said how are we going to control this right?

[00:56:06]And there wasn't like a million ideas.

[00:56:08]There was I mean there may have been but most of people were like you know what we should do is if like go from just we'll pay everything to we'll have a fixed amount and if people want to go to the expensive steakhouse they can pay for it themselves. You mean they weren't like funding Ubers to take people from the steakhouse to the lowerc cost restaurant and like you know cover >> Yeah. and like a sea of of of like restaurant connoisseurs to like call people and say you know there's a different steakhouse you could go to maybe would you like to like all of these things they weren't doing an all you can eat trying to go find the restaurants and doing an all you can eat buffet like none of these things make any sense lead back to Denny's >> whatsoever like whatsoever. there would only be one solution which is to say like hey you can go wherever you want but like if you go here basically it'll be basically free it's really high quality it's awesome it's a great meal it's great um and if you want to go somewhere else where by the way in healthcare you're going to get way worse food you know treatment whatever that's on you that's I think the way I view it again I'm biased there's only one solution it is the only solution it has to be the only one which is a change in the expense policy and everything else would be sort of categorically absurd um as a way to handle that problem because the essence of the problem is an economic and a purchasing one not anything else and so that would be my view and I think it's re if you just do that one change you solve US healthcare like overnight um >> reason they haven't done that right this like it it sounds very simple when you say it like this right but like >> it's not like healthcare expenses are a new thing for employers right now so why haven't they done something like this if it's if it's so well I think it's it's two or three things. Um the first is that um these programs and it and all sit in they they sit in the benefits or um where the goals expressly and I think totally reasonably for many organizations have not been cost optimization. They sit in the benefits world where we want to make sure we are taking care of our people and delivering high quality care. um even then again I think you there's a reason to really make sure people are going to the highest quality providers but um it hasn't been a priority to solve this and and it I think expressly in the vast majority of organizations um it if it was sat under the CFOs like they would go figure this stuff out the same way that the expense reimbursement policies sit under the CFOs and so I think mostly um it hasn't been that big a deal and I think people have been very much fine to say listen you can go wherever you want.

[00:58:43]Like if you get really sick, we want to make sure you can go wherever you want.

[00:58:47]And it doesn't matter what the cost is.

[00:58:49]Um and as soon as it matters what the cost is. And by the way, the nice thing in healthcare is you can get high quality and low cost. But um as soon as cost becomes a part of the equation, there's a very clear path to solve it.

[00:59:00]Again, I think most organizations just aren't there yet.

[00:59:03]>> Yeah. It feels like there hasn't been this internalization of basically that like all this increase in healthare costs is just suppressing wage growth, right? And then ultimately like you literally are just trading them off. and they sit in different again I think that organizational design decision as a CEO like matters a lot like in Garner you know who runs our health plan is our CFO um and so like he's responsible he makes the calls like that's what it is and I think if and again I I'm not advocating that for every company um but I am saying that that it's a fascinating decision where the financial mindedness and then the care-minded are split in most companies and of course that creates the outcomes that we have today.

[00:59:39]>> Yeah. Well, Nick, this has been a fascinating conversation. I'm honestly like one of my favorite parts of of having the podcast is just that more folks get to hear from you on this stuff because uh I always really enjoy our conversations. I want to leave the last word to you. Like where can our listeners go to learn more about Garner?

[00:59:51]Uh the exciting work you're doing.

[00:59:53]Anything you want to point them to? Uh the mic is uh is yours.

[00:59:56]>> Sure. Yeah. Um so you can check us out on um on garneralth.com, check us out on LinkedIn. Um and um we'd love to hear from everybody. So, um, whether it's the plan for a job or interesting about some of our research there, you can get some of the, uh, follow-ups there. So, >> yeah, and I think you have some, uh, if I recall when we're releasing this, you'll have some exciting news to to announce on the company side, right? Uh, what's what's the latest and greatest?

[01:00:19]>> So, we, um, have announced that we're doing our series E, uh, with Index Ventures. Um, really grateful to the team there. Um, values the company at at, uh, just around 2.7 something billion dollars. um a reasonably large step up from last round. So, we're really excited about that. We got um I think really good partners in Janvi and Martan and others at Index and um we think this is going to really be a a good step forward for the business and and we're really excited about the the next leg of growth here at Garner.

[01:00:50]>> Yeah. Well, we couldn't be more excited too and and thanks again for just an amazing partnership these last five years.

[01:00:54]>> Yeah. Thank you both. Thanks for having me.

[01:00:55]>> Awesome. Thanks for coming.