September 15, 2023
Welcome to Perspectives, Leerink Partners’ signature podcast, where we share our insights and interview leaders across the industry to get their perspective on how they’re driving innovation. We’ll also be digging into the backstory to learn more about what has most influenced their success. Be sure to check out all episodes by Leerink Partners.
Thad Davis: Hi, how’s everybody doing? I’m Thad Davis. I’m a senior managing director at Leerink Partners. It’s my pleasure to be joined here by Justin, who’s the founder and CEO of Sift Healthcare. So, a guy I’ve known for a few years, through a couple of fundraises, et cetera. Isn’t that right, Justin? I think we met a few years ago through a mutual friend, Jason Helgerson, if I recall.
Justin Nicols: Pre COVID in Milwaukee. Yeah, you were here.
Thad Davis: Oh yeah, pre COVID, yeah, in Milwaukee. I was with Jason who, I think I’m, I think Jason’s going to be on the podcast here soon, so we’ll have to get him on here and have him talk about like value-based care and how good it’s to be an independent consultant.
Justin Nicols: Quite the evangelizer. Yep.
Thad Davis: Exactly. Hey, before we begin, just to try to set, do a little bit of level setting, since, you know, people, we have a lot of different, founders and executives on the, on the podcast here. Maybe just give us, we’ll come, we’ll start with kind of, it’d be great if you could just give us an overview of what Sift is, and then talk about what you did, what you formed, talk about what the company does, so people understand kind of what you’re into on a day-to-day basis. Then we’ll, we’ll take a step back from that, you know, travel your journey, up to that point and then that’ll lead back up to we’ll come back and hit more deeply on Sift. But maybe just give the audience a little taste of what you’re all about at Sift right now.
Justin Nicols: Yeah, no, it’s great to be here. Thanks Thad. Yeah, so Sift healthcare five-year-old business data and analytics focused on healthcare payments. Primarily two parts of the business that do coincide, patient reimbursement and payer reimbursement, really optimizing, enabling patients to pay flexible payment plans. So that involves the right communication, the right payment plan strategy, increasingly with price transparency and consumers have been no surprise billing, patients rightfully so expect a better billing experience, right? And we’re supporting that through advanced analytics on the payer reimbursement side, really working with large healthcare systems to combine clinical data with payment data and predict and intervene potential adverse payment outcomes sooner, while ideally care is still being rendered. So really, at our core, we’re a team of data scientists that are very, very good at data standardization and normalization. And principally speaking, what that means in healthcare is really trying to knock down data silos between clinical data and kind of traditional back-end claims data and making those data sets talk to each other and communicate. So, we go from being, more preventative rather than reactive and kind of these negative things you experience in revenue cycle management.
Thad Davis: A very simple task that has been solved many times.
Justin Nicols: For years, exactly.
Thad Davis: I’m working on a problem that’s already been solved the uh, Yeah,
Justin Nicols: It’s amazing as you know, I mean you look at the stats and it’s kind of good for us to be grounded I mean you look at you know, healthcare is about 17 percent of GDP and this administrative side is 20% of that, so you’re talking nearly 4% of GDP is spent principally trying to work out payment between either a payer and a patient and It really comes down to data complexity in our viewpoint.
Thad Davis: That’s kind of the interesting thing. It’s like a you’ve formed a business that’s basically here to fight a payment and a payment bureaucracy that effectively exists and try to, you know, stream, begin to streamline, take that cost out. I mean, it’s just such, there’s just so much waste. It’s like, yeah, like revenue cycles, one, you got to get the money into the organizations, of course, because the reimbursement dynamic, but it’s also doing it efficiently to avoid that, that massive amount of waste that every organization experiences, but tends to grow unchecked, you know.
Justin Nicols: To that point. I mean, it amazes us, it amazes us. Like, we understand that a certain level of work needs to take place to get dollars in the door, but when you see to your point, wasted work on dollars that are never going to come in the door, that’s where it just becomes this vicious feedback loop of wasted dollars on dollars, chasing things that’ll never get paid.
Thad Davis: Misallocating, misallocating time. You’re like, “why, why don’t even allocate any time to that? It’s never going to get paid for that. You’re like, well. Sorry.” Yeah. So, you’re trying to solve that, solve that thing. So, you basically, you came up with this idea in middle school and then you had to wait for the, wait for the industry to move towards you. So, is that, is that how it began?
Justin Nicols: You know, it’s a, it’s a, it’s, I, I am a, as you know, I’m a healthcare outsider. So, I started looking at healthcare payments more broadly probably seven years ago and, my first kind of gut check at, attacking receivables more broadly was, was really an idea around mobile payments. So again, seven years ago, really saw the power of mobile payments and, and really thought that was going to dramatically affect an impact to people’s ability to pay, but then also how is that going to affect businesses ability to collect payment right? That could kind of Receivables management more generally so the actual iteration that first started sift was it was a company called “remind a bill” so remind you about your bill and I was an early adopter of Twilio or Stripe, and Stripe, this again, was probably six and a half, seven years ago, and you could go into a dashboard and automate receivables follow up. So, tone and frequency of contact and medium. So, text, email, call, and then how aggressive you wanted to be, and kind of verbiage, right? And it was a good idea, but, but very quickly, you know, I realized being a first-time entrepreneur in Wisconsin, having never raised money, having not gone to a pedigree university, it was going to be really hard to raise a 10-million-dollar seed round to sell 100 dollar a month subscriptions to small businesses. So, I quickly understood that. I was going to need to go vertical specific, and my criteria was high cost to collect, high bad debt write off, non-recurring payment. As you know, you’ll only find so many things, utilities a little bit.
Thad Davis: I know a space for that, yeah.
Justin Nicols: So, started to find, uh, obviously healthcare. And I was still focused on really the mobile payments component for patient responsibility. There’s been a lot of companies that have raised money focusing on that and did that in a parallel path as I continued to explore, the space and very quickly realizing there’s nothing against those companies, but at its core merchant processing in an iframe from mobile payment, isn’t that sophisticated of technology for better or for worse. I, as an entrepreneur, I get drawn into complex problems and it continued to blow me away the level and amount of effort that people were exerting to collect money when in my dumbed down outsider brain, it was the same two repeatable transactions, either the payer, the patient, like if you, if you’ve never heard of this thing called machine learning, get all that data in one spot and you could probably do some interesting predictive analytics. So, at that point I was naive enough to think you could start a data and analytics company focused on healthcare payments in Wisconsin.
Thad Davis: That’s actually, I think, an interesting, yeah, exactly, like, you know, “like, oh, this, I’ve achieved, I have the high mountain, so I just need to climb it here in Wisconsin.” Going back, you made a comment about, you know, the pedigree university, and I mean, that, that’s been, that was one of the things I had here in my notes. I mean, like, you’ve kind of, you came out, you came out of, you know, college, and then you did a couple of different things behind it, behind that. Like, like how did you, like, were you always like, you know, “I came out of college and I was like, listen, I’m going to have to hustle” and not to make this a show about Milwaukee, but the, “I’m in Milwaukee, I’m going to have to hustle.” Not that, not that Milwaukee’s got a lot going on, but the, but I mean, how, how did you kind of go through this, the thing between like PS, Brady, Lean Tech? Like how’d you kind of connect up those dots there?
Justin Nicols: Yeah, so I mean, I was never a really good student. so, I was never gonna get into a good school.
Thad Davis: I was never a good student either. So, yeah, like, I’m like, I’m a, I’m the, uh, high school delinquent done well here.
Justin Nicols: Yeah, exactly. So, I’ve always been drawn to capitalism from a young age, and I’ve probably have entrepreneurial leadings, over the years, looking back, I tried to start things, my first thing I think I started in 6th grade, I had an original domain called RentMilwaukee.com, which was, this had to be, you know, 1996, something like that.
Thad Davis: You’re like a domain troll.
Justin Nicols: Well, if I would have kept it exactly. Now I think about that. Cause I had an uncle at a couple of bucks who bought me the domain. I didn’t have a website. Obviously, these websites were crazy to get built there in terms of cost. Right? But I would call, I didn’t really know what a business model was. I would call limo companies primarily and I would say, do you want to advertise my nonexistent website? And I would try to negotiate for rides to school. Cause I thought that would be cool. No one ever took me up on it, but I banged the yellow pages for a bit. So I’ve always been drawn to capitalism. and I initially thought maybe my angle was probably a more traditional finance angle, such as yours. I sat for level one CFA. I interned at a, you know, a small middle market, private equity shop here in Milwaukee and very quickly realized, you know, this just isn’t going to be a game, I’m already disadvantaged and for better, for worse, as you know, I mean, you’ve had a great career in finance and continue to like, it’s a pedigree world for the most part. And it’s a relationship thing. And if you’re trying to crack into that from an outsider, it’s going to be tremendously difficult. And so again, always kind of having just this for better, for worse hustle mentality and thinking that I can sell a fair amount of things through conversation. I, I early in my life, my mom, who had great influence on my life, had business cards made for me and my type, my business was talking, talk people into things. I, um, so, I’ve just always had a knack and been drawn to kind of complex things and trying to reinvent them for better for worse and, you know, hopefully make some money along the way, but certainly I don’t think, um, any other path for life probably wouldn’t have worked out for me. I think this was defined.
Thad Davis: Are you the type of person where you like see a process and you’re like “That is so messed up. I have to fix that.” Are you, are you, do you like, do you like process problems bother you, you’re that type of person?
Justin Nicols: No, I mean, so I think, um, disjointed, yeah, certainly. Like, I mean, there’s always process has its role. I mean, there’s a decent book that I think had an influence on me, and it’s a bit dated now, but it’s called Eat People. And it’s, it’s, it’s principal, right? I mean, you’ve heard like “software is going to eat the world.” So, I mean, I’ve been, before it was cool to say things like “data is a new oil”. I’ve always been a firm believer that, you know, technology is going to have this interesting effect on us. And, data is probably going to play a major component of that. So previous to Sift, I was involved with a venture backed startup, that was trying to do predictive analytics around consumer purchasing insights. So, this was seven or eight years ago. So far more interesting then but, trying to arbitrage on that ad inventory, so could you predict what that buyer would like and then serve them up affiliate marketing?
Thad Davis: Yeah, the ad tech market’s hardcore, man. That’s like, once you get into consumer preference, that’s like a hardcore, I mean, not that revenue cycle is not hardcore, but that’s like, that’s like really hardcore.
Justin Nicols: Yeah, no but things, I mean, simple machine learning has been figured out, I mean, Google’s had this thing figured out for, you know, 20 plus years now, right?
Thad Davis: Right.
Justin Nicols: That’s the viewpoint I started bringing, for better or for worse, to Revenue Cycle was like, okay, data management platforms have existed forever in ad tech, as you know, compared to healthcare. And machine learning, optimizing conversion in ad tech is like, if you’re not leveraging some type of ML to optimize conversion, you’re an idiot in ad tech, right? And so, for better or for worse, that, that influenced me in terms of seeing the potential of what ML could do, but then probably it also helped because it was so well defined in ad tech that it helped me be naive to think it’d be simple in healthcare, right? Like, and, and it’s not, I mean, looking back now when I’m almost five and a half, six years into this and having raised 15 million dollars, like would I have, if you could have said, “I’m going to be here in five years and here’s everything you need to go through to get there”, I probably would have thought you’re crazy, this is going to take me 18 months to, you know, hit stride and really have this thing go gangbusters. And, and as you know, things in healthcare just, just take a lot longer.
Thad Davis: As you and I have talked about that, like, I think the space, I mean, you’ve been successful kind of getting execution done, and I think that there’s a lot of, there’s a lot of companies now that are going through that phase where there’s like, you know, healthcare. I mean, I think like, uh, one of the big venture capital firms published an article recently that said, “healthcare is hard”. And you’re like, yeah, healthcare is hard because people think about walking in. I mean, I mean, but on the other balance of that is what you said, like the naivete of coming in as an entrepreneur, especially a non-healthcare entrepreneur into the space here. Like, I mean, even myself, I would be daunted to be like, I’ll make a revenue cycle. I’m like, “oh my word. I’m going to spend all, like spend a few years looking at that. Oh wait. Justin has spent a few years looking at that.” I mean, I mean, that’s, that’s I guess, precept number one for, you know, outside in entrepreneurship. Don’t think about it too much.
Justin Nicols: That’s exactly right. I mean, it’s just, it was mission bound. I mean, it, it, it became, I was very determined to make it work. Like, I mean, looking back now, I’ve got a 15-month-old daughter. Like, could I even do the things I did five years ago? And I, I mean, when I left out, when I was at this other venture backed company, I brought an idea to the board that, um, that would have somewhat disarmed the founder who wasn’t all that effective at managing the business and, and would have required about 3 million dollars more of investment. It was probably the right idea, but I was the wrong messenger for it. So we ended up parting ways. So, at that point it put a chip on my shoulders. Like, well, if you’re not going to let me help influence in this company, I’m going to prove it out somewhere else, which you need to have a certain amount of grit in rev cycle, no matter where you’re from, where you went to school, who your investors are. Couple that with, barely graduating college, being in Wisconsin and raising, you know, my first round, we’ve raised 15 million now. My first round was a 660, 000-dollar round, right? So, like, you can’t even hire, you can barely hire a data scientist, get cyber insurance, and spit up AWS for 600,000 dollars anymore.
Thad Davis: That’s right, yeah.
Justin Nicols: So, you need to kind of be somewhat dumb or oblivious to how hard it’s going to be to actually do that undertaking, I think.
MUSIC TRANSITION
Thad Davis: What about, actually, that’s another interesting vector here that I’ve written down. It’s like, I’ve met some of you, I’ve had the fortune of meeting some of your other team members, but you’re not a classic data guy. Meaning like a scientist or somebody, somebody that’s done like, you know, linear algebra 1, 2, 3, 4, 5. How has it been for you and how do you find and unite with these, with, with data scientists? I mean, this is, they are also a very, very specific type of mind, very specific type of individual and have a tremendous amount of demand. And especially a couple with somebody like you who’s not a natural, like, engineer, so to speak.
Justin Nicols: No, I have no technical background, right? And I think to add to your point too, like, risk averseness, typically through the roof for a data scientist, right? Like, I mean they calculate risk pretty effectively. So, getting them to work at a startup is, is difficult. I, I think, you know, again, I haven’t achieved much, but my ability is frameworks, right? So, seeing what data has been done in FinTech or AdTech, right? Like being able to communicate that to an opportunity to a different industry and making that digestible, is probably what I see and I’m able to communicate. So then it’s a matter of. So, in early on, I mean, I would send hundreds of LinkedIn messages a week, which is actually how I got to Jason through the Wisconsin network, but that’s how I got to Jason initially, as anyone that I thought could understand rev cycle and talk to me about it, I would try to talk to. And so, my number one job was keep the company funded and then find partners that were willing to innovate. Like customers, channel partners, et cetera. And I think being able to communicate the potential through a simple framework and then painting the opportunity to the industry is what excites data scientists if you can do it in a pragmatic way where you’re giving them enough of a cushion and security in how they work.
Thad Davis: Yeah. Identifiable problem. Identifiable commercial need. Opportunity to do something. Create. And then, you know, kind of, it has to be real. Yeah, I have a couple other clients on the data side that are like, they, you know, people used to go like, they would be like, “Oh, that’s exploratory in the data world.” Like, “I’m going to have to do literally, primary data research to see if that even has a trend in there” versus like, “We know what the problem is, here’s what the data is, it needs to be solved” and you’re like, yeah, it’s like solving a math equation, it’s like the equation can be solved, you have to solve it and then, you know, somebody actually, you try to find somebody that’s going to get excited about that.
Justin Nicols: Exactly right. And the right people do, right? And those are the people where you want, I mean, they think at night, all the time, they lay in bed trying to solve the problem, right? I mean, these are naturally very inquisitive people that, it almost disrupts their life if they don’t get an answer. And that’s probably good and bad for them personally, but if you present an opportunity to them and they buy in.
Thad Davis: It’s like a movie. I have to know.
Justin Nicols: We’ve got one data scientist for us. We’ve been missing and it’s not even a core data element anymore. And I, and I love this person. She’ll bring it up every six months to me and I’m like, “I’m not going back to the client and asking for that.” Like, we don’t need it. It’s gonna take a long time. I have to, yeah. It’s like bothering me. They need that piece of the data. They need that answer in their life and you know, that’s what makes them good.
Thad Davis: They’re writing it on the walls of their apartment. Yeah, “Like just one piece. I won’t be fulfilled until that one piece of data is mine.”
Justin Nicols: Yeah, exactly. Somewhat how they are.
Thad Davis: I mean, how do you see like transitioning, And I mean, I mean, uh, over the last few years, I know that like Sift has started out in one direction. I mean, it is always directional changes in customers and clients drive that, but the space is transformed around you as well kind of take you just take a minute and kind of talk about the journey from like Sift one to Sift five or whatever, however, you want to version it there?
Justin Nicols: Yeah You bring up a really interesting point that we could probably talk quite a bit about and I think to your point it’s part of the industry transformation, and people likely becoming more aware of what actual problems are, rather than this huge machine doesn’t work.
Thad Davis: It just sounds funny. It just sounds funny. Like people are becoming more aware of actually what a problem is.
Justin Nicols: I mean, all these things is that machine learning is going to start. Solve everything. It’s like, well, RPA didn’t work. Like what’s our actual problem? Like, check that one off the box. Like, it’s like, well, it’s probably ’cause your data doesn’t talk. Right?
Thad Davis: That’s why I say to clients, like, I’m on a, I’m on a call with them. We’re talking about some like fundraising thing. We’re like, “Have you considered using generative ai?” They’re like, “What’s that gonna do for me? Absolutely nothing. That’s why I’m, that’s why I’m here.” Yeah, and so you’re like, “Really? Could I use generative ai?” “No, you cannot use generative AI no.”
Justin Nicols: And it costs you 30 million to build the model. So, for us, I mean, literally, I mean, the first criteria for us to get was high-cost neglect, high bad debt write off, non-recurring payment to do data science. I always you need data, right? So, we had to find a big enough client. And just through our network, we got to know the early out world, right? Day zero to day 120 patient billing. And I think for us, we began to understand that the true value prop is not, and as you know, the space is littered with services companies and rev cycle, and that’s been primarily a component of labor arbitrage through offshoring or better management of labor staffs.
Thad Davis: You have to be on the value, that’s the problem is it’s commoditized quickly, you have to be on the value side. If you’re simply, you know, you’re just doing like, you know, uh, like manual recoding work, you’re like, “Mm, yeah, like, yeah.”
Justin Nicols: So, we quickly realize while these outsourced and there’s great ones that we have great partners, particularly one that we love, it’s commoditized, right? And so, if you look at where’s the true value lie, the true value lies where the revenue is earned and generated, i.e., the provider, right? So, you need to work directly with the provider. And I also think the industry has again, shifted from “How am I reactive to an adverse problem that happened, i.e., a claim got denied or this patient hasn’t paid in 60 days” to “How do you rather be more proactive and get in front of problems before they get created, right?” And so that’s where we talked about earlier about just bad, human capital chasing bad money. You’re never going to collect. There’s so much of that. It’s, it’s how do we get this information upstream sooner in the process, so you can mitigate a negative payment outcome from occurring. And if you can’t just know that so you’re not exerting resources to try to collect it and just chasing bad money after bad money.
Thad Davis: This understanding that you’ve achieved, which is a sophisticated understanding that, that was built over time. That was never day one.
Justin Nicols: No, I mean, it’s funny, like, so now we, I mean, our principal goal, particularly on payer reimbursement is to make clinical data sets talk to payment data sets, and then create action in clinical workflows. Very complex undertaking. We would get in meetings. Our, our head of analytics used to be VP of insurance claims analytics at a leading data company. And they had a contributory database that had clinical data, all types of data, pharma data, all types of, and he would always say we’d be, we’re going to want the clinical data at some point. And I never, it took me two years to figure out how clinical data, which sounds so dumb and naive, how clinical data influences reimbursement, right? Which is principally how the bill gets created, but I could never understand. Like, I would say, “don’t say that in a meeting because the meeting, the second you say that you’re going to scare them”, first of all, because, you know, clinical data becomes much more guarded from a security perspective, which it should. But then also we’re going to have real use cases, right? And you start talking about touching clinical workflows. Now you’re getting a whole new thing that you’re in deep, deep waters.
MUSIC TRANSITION
Thad Davis: I understand the comment you just made, but isn’t that, that’s the battlefront, right? Because there’s everybody, you’d be surprised. It’s like one of the things you’d be surprised. Everybody knows what the problem is getting people to adopt the solution is different. They’re like, “well, the solution is unifying it with clinical data because then you can get ahead of the problem.” They’re like, people are like, “Whoa, whoa, whoa, whoa, whoa.” But that’s the solution. They’re like, and it takes them, you’re like, “Don’t ever say that in a meeting because we need to wait two years and then I will say it in a meeting and then you’ll know it, the time is here.” And it’s, yeah, it’s like, nobody’s going to buy that. That might be right. But nobody’s buying that.
Justin Nicols: It’s a sacred thing is, you know, I mean, and again, clinicians should be sacred, but at the end of the day, if we’re principally talking about how to avoid negative payment outcomes. Like to your point, like it’s got to be well care still being rendered. Right? And so, for us, the way we’ve done that is through partnerships of existing industry leaders that have that workflow access that are trusted resources in driving clinical workflows because if, it would take a company like our size years to break down clinical workflows and get trust there, right? It’s how do you partner to drive and add value to the existing process, because going direct and just saying, I’m going to introduce a brand-new clinical workflow. I’m a startup trust my machine learning models. No CMO is ever going to buy that.
Thad Davis: There is, more importantly, that decision making chain is like elaborate at that point. You have to, you have to have an empowered, you have to have somebody like a CMO that’s empowered. More importantly, because they’re making a fix, not necessarily for clinical utilization, but for revenue utilization. They’re like, “What’s this doing to my clinical?” Absolutely nothing. It just helps us build. They’re like “Wait one second.” Yeah, they think they tend to think, but now, now that’s more de rigueur back, you know, five, seven years ago, they were like, “What?”
Justin Nicols: I would say the beautiful thing. I mean, you touched on an interesting point that we’re knee deep in right now is obviously clinical burnout. You’re seeing some really cool things going on actually with generative AI around notes and dictation. But for us, the interesting thing, because the way you get by into the point that you touched on is all of these engines, because they are siloed data sets, they’re clinical purist engines, so they want to know the maximum amount of information to justify whatever decision that engine’s making i.e. you know, what DRG are we recommending or how are we coding this thing? Right? And the way you start to get buy-in from clinical leaders, because they know this, is if you can help have the payment data inform those clinical workflows, significant and material amount of those nudges and queries that result in those clinical leaders saying they’re experiencing burnout are actually wasted. Right? Because if you know you’ve gathered enough information with a certain degree of certainty, that you are going to get reimbursed in this thing so you don’t need to beef up the note or whatever that you don’t need to query the physician anymore, then you should stop. But these engines don’t understand that. So, they’re clinical engines. They want maximum amount of information. The same is when you’re talking about bad money, chasing bad money, if that thing’s never going to get paid, that insurance is never going to cover that thing. Then we should know that sooner. So, then you stop querying, right? Stop, stop the clinical burnout that results in that. So can we actually narrow the scope of focus in what those clinical workflows drive from a querying and nudge perspective, you can actually drive up an improved financial impact while hopefully reducing the clinical burnout and making the clinician feel better about what they’re doing and what they’re focusing on.
Thad Davis: Yeah, that drives you down like the, the road of, you see a lot of discussions in the space around like, you know, and not discussions, but you see hyper-specific revenue cycle products and things like that. Like, Dermatology. Let’s pick one out. Like, oh, it’s just this thing is just a DERM workflow because this has to be united clinically, it has all these preventions, we can get better collections. I’m not sure I’ve seen like a great analysis to say, oh, this hyper specific product has a hyper specifically better outcome on revenue cycle collection. But that’s where people like that’s what, that’s a clinical back fit going on there.
Justin Nicols: No, that’s exactly right. And I think what’s interesting I mean, you touched on an interesting topic that I’m sure a lot of your clients and friends who may like the interesting thing where those things have been deployed effectively are typically private equity owned, right? Smaller workflow, smaller, narrow focus, where they are willing to invest in technology, right? I mean, I think when you start looking at larger acute settings, like you said, it’s been really difficult to crack that clinical workflow not to try to reinvent them to attack revenue cycle problems because for better, for worse that that dollar incentive hasn’t been there for them. I think increasingly with margin pressure that everyone’s becoming sensitive to that. And I think again, if we can couple that with how do we reduce nudges and queries and actually reduce clinical burnout and optimize payment outcomes, then you, you, you hopefully find a sweet spot.
Thad Davis: How about just, uh, switching gears for a second, how about, like, let’s go back and revisit a conversation you and I had a laugh about a few years ago, and that’s just the proliferation of the term “AI”. I mean, like, versus like, the difference between learning, the true learning algorithms versus like just a really big algorithm. what’s your current thoughts on this, this, I mean, especially now more than ever, I mean, we just put generative Sift and we’ll go raise you 100 million dollars.
Justin Nicols: Um, I think is, you know, there’s always been rule systems that are fancy algorithms that I think a lot of people have tried to masquerade as quote unquote AI.
Thad Davis: The amount of AI companies where I’ve called them, they never do this as an investment banker. You go out and you call an AI company, “A big adjudication” and they’re just like, “What did you just call me?” I’m like, “You just, you’re like an adjudication. You’re just making a set of decisions based on a set of rules. They’re like, “We’re way more than that.” Yeah. I was like, “I don’t think so. I don’t think so.”
Justin Nicols: And I, and again, dumbed, like to your point, I mean, dumbed down way I understand these things is like, well, like what data talks to each other to learn, right? Like that’s the simplest way to understand machine learning or feedback loops. It’s like, okay, you make a recommendation and prediction. How do you quantify what the desired outcome was, and then what was the actual outcome? And then how do you drive either the positive or negative back up to the front of the engine, right? And it’s like, if you’re not doing that, like that’s there’s supervised learning, unsupervised learning, it’s not learning.
Thad Davis: If it’s just sophisticated Plinko, it’s not going to, it’s not going to be, it’s not going to work in that way. Yeah. It’s that feedback, that learning algorithm, where it’s like the outcome, you wanted this, this occurred, why did it occur and how do we change, how do we change the beginning to prevent the end? Otherwise, if it just does the same thing every time, you’re like, it’s not generative AI. It’s just recording it.
Justin Nicols: No, you’re not learning. There’s no feedback loops.
Thad Davis: And it’s not generative. It’s not AI.
Justin Nicols: It’s just not AI at all. Exactly. And that’s, I mean, that’s where. So, the complexity of reimbursement has never been greater between the payer and provider dynamic. It’s a living, breathing thing. You’re seeing what in our viewpoint is kind of the creation you touched on, kind of an adjudication engine like clearing house 2.0 now, where as, you know, years ago you used just to send a claim and a claim would get paid or not paid. Now you send a claim and send me that clinical documentation, right? So now you’re talking about structured or unstructured data along with the structured data of a claim. So, what are they seeing in that structured or unstructured data of the documentation that drives reimbursement likelihood. It’s very difficult to figure out at scale unless you’re leveraging advanced data science platforms that are doing real machine learning. Say, you used to get paid this way for this reason. They’ve changed this and are saying they’re not paying it for this reason. This is how you start learning from that to drive those learnings upstream. I think it’s, it’s, it’s somewhat bad for the industry and bad for society, but the complexity of reimbursement is probably not going anywhere. So, you need these feedback loops and you need a way to understand complex data sets between structured and unstructured things. And if you can knock down those walls, then you, yeah, then you’re doing machine learning, but really all you’re doing is making things talk so you can learn from outcomes.
Thad Davis: That’s right. Yeah, so it’s like, well, I had 25 of these rejected. Why? Here’s the commonality. Okay, don’t ever do that again. Or, or no, or go, or yeah, or no, or go, go tell doc, you need to start collecting this. You’re like, oh, okay, will do.
Justin Nicols: Or no, right? I mean, there’s different ways this gets attacked, right? Like, okay, does that need to be taken into account in contract renegotiations between the payer? Or are these things that, yep, we see why the payer’s not paying them, but we feel like we need to do for the patient, then just know it, right? But there’s so much spent cost on the back-end chasing things that are never going to get paid. How do you know these things sooner?
Thad Davis: How do you not get sucked down the rabbit hole of just do not doing just straight services around the, eventually you’ve, I mean, you’ve acquired so much knowledge and the team and the firms acquired so much knowledge. It’s like, “Oh, okay, we have these products, but effectively we’re just doing consulting work for you. Yeah, we’re just going to bring on some consultants. How about we just do, then that transforms into how about we just do it for you?” How do you avoid that, that, the services rabbit hole? Or do you?
Justin Nicols: No, no. I think it’s, I, you raise a really interesting topic that I don’t, and I mean, there’s been great firms out there that have, venture firms, that have written, like there’s very limited peer data AI platforms in healthcare that can exist without a certain element of services.
Thad Davis: That’s very true.
Justin Nicols: And if you’re saying you are, you’re masquerading, burying the surfaces somewhere else, right? Because our viewpoint is machine learning can do great things, right. But really at its principal basis. We’re recommending the next best action in the existing workflow and how that gets done. Yes, it’s through machine learning and taking complex and disparate data sets, making them talk in the background and then driving real time thinking into the existing workflow. But so much of healthcare is change management, right? Like it’s thinking and actually what you’re doing upstream needs to be unlocked through conversations. So, we see a healthy balance candidly of where we actually, we build reports and for better, for worse, we pride ourselves on this as a quote unquote AI company, like we provide quarterly ROI, hard dollar value of our platform, where have we performed, but on a monthly basis we provide subject matter expert curated reports for CDI, coding, auth, infusion, back-end denials. Where we’re identifying more change management things that technology can’t fix that the boots on the ground need to be understanding from a process and workflow perspective.
Thad Davis: Yeah “We’re doing this for you. We’re producing results. But as we’re doing this, we’re seeing the following things happen. If you would do the following things. Yeah, we’re, we’re here to, this is our value add on that basis.” Yeah. And then, then again, then it’s like, “Hey, could you just do that for us?”
Justin Nicols: That’s the simply, so wait and our, I mean, talk to me in two years as I continue to, you know, traipse around this space, we’ll unlock insights. We’ll never do the work for you. Right? Like, so I’ll drop them on your desk. Show me the worker, but I’m not going to be the one who, who does the work. Right? I think there’s plenty of firms out there that have created this labor arbitrage and we’ve got some great partners there, but for us, it’s really about unlocking the data, making the data talk, drive the next best action in the workflow and empowering the workers with intelligence.
Thad Davis: All right, I’ll hit the topic. How about Milwaukee? Like, you’re like, you’re my, you’re my healthcare entrepreneur client in Milwaukee. I mean, the scene up there, it’s actually, I think, as many people probably listen or know me, know that, yeah, I used to live in Chicago. And I think that, like, once you get into the community and the entrepreneur community with these, those, the city’s that corridor is a generally risk adverse, risk adverse area, heavy industrial focus, just traditionally, and then, but there is a real startup community that flows through the quarter there, and it’s been building up over the years with some local successes and some non-healthcare and some healthcare sectors. But what’s it been like doing all this in, especially in healthcare in Milwaukee?
Justin Nicols: You touched on some interesting points. I mean, clearly Epic in Wisconsin, that helps us a bit, particularly from a headcount perspective, right? Like, I mean, we find really good resources. Nordic was obviously, you know, one of the bigger Epic implementation shops started in Madison. So, there’s a lot of obviously epic expertise in Wisconsin. The problem to your point is the risk averseness that venture firm that you mentioned earlier that wrote kind of the thesis on like “healthcare is hard. Startups are hard here.” Like I think their horizon they gave on average to get to a hundred million in revenue was 10 years. And so, which is a really good outcome for a technology company in healthcare, right? You’re like, that’s probably likely, you know, half a billion, right? Like it’s a good outcome. So, take that 10 year time horizon and then communicate that to an investor who a, doesn’t understand healthcare is barely done venture and then you tell them it’s gonna be a 10 year ride That’s not the most compelling deal that they’ve probably seen at their desk, right? So I think it’s getting better here. We’ve been a great beneficiary of kind of state sponsored venture capital here for the early money that we got where the states kind of start these master funds and then seed sub funds. I would say, you know, out of the 15 million, you know, probably eight is in state. You got seven out of state, but our first six.
Thad Davis: That’s pretty good. That’s pretty good, actually.
Justin Nicols: Yeah, but our first six was all in state, right? It was my network, who I know and knew. But I mean, there’s, it’s one of these things, I think, chicken and egg, where there’s a decent amount of companies that have been started. I think they are somewhat capital starved comparatively on kind of a national basis, but now it’s, it’s, it’s people like me’s job to deliver exits so then you have shareholder return to hopefully kind of change the risk averseness where people take more risk, invest more of these deals and hopefully more funds get created. Right? I mean, there’s just, there’s a handful of a hundred million dollar plus funds, which is laughable in this space is, you know, I mean, you need big, big balance sheets to support these activities. So I’d say it’s getting better. But there’s still not enough capital being deployed. But again, what’s the argument from the entrepreneur? Then it’s like create exits, I guess. And so, focusing on the thing we can control is probably the best thing, but there’s certainly areas you can raise money easier.
Thad Davis: Also, I think my belief and you can agree or dispel this, but it has been that in, in, in the, and especially in the Midwest, upper Midwest ecosystem. Once you’re actually in the system, it’s actually pretty collegial. Like, meaning you’re funded, you have something going on. It’s a small community, it’s collegial, it’s not, you would want it to be bigger. On the other hand, you don’t have maybe some of the, maybe some of the, The, noise or the, the, some of the coastal, like I, you’re just another, you’re just another portfolio company, like move on, you know, yeah, it’s, it’s a little bit more, uh, familial, so to speak.
Justin Nicols: Totally know what you mean. We see the majority of our investors still roll up their sleeves, like to your point, they’re trying to make introductions for us and really understand the business and yeah, to your point, would you rather have someone with a more powerful Rolodex, but you’re only getting 10% of their thinking, or would you rather have someone who has a less powerful Rolodex, but is rolling up their sleeves 70, 80% of the time for you? So, we certainly, and a lot of these, particularly our investors, you know, you’re talking the most advanced fund that’s invested in us is on fund three only, right? So, I mean, these guys still need to prove themselves too. So, you’re seeing that kind of gumption of “How do we roll up our sleeves? How do we create exits? How do we empower our portfolio companies and make sure that we’re creating the relationships, with other funds that can support the next life or the next phase, the next round.” And that’s how we got our last round done was one of our existing investors. That’s a 30-million-dollar fund, small, right? Made the introduction to 150-million-dollar fund, right? So, those relationships are so key. And again, that’s why I feel like We’re going to figure these things out over the next, you know, two to five years. We’re like, hopefully some exits get put up on the board. We make some people look smart, make some dollars, and hopefully there’s more capital to get dispersed.
Thad Davis: Absolutely. The town, the town’s great. I mean, the town’s great. Very supportive.
Justin Nicols: There’s great universities. I mean, that’s, I mean, you look, I mean, there’s Marquette, MSOE, you know, Wisconsin, obviously, I mean, Epic’s here, right? I mean, there’s, there’s so many, uh, but I mean, kids leave. I mean, there’s not a lot of, you know, there’s not enough, I mean, particularly Milwaukee, you know, it’s like Harley Davidson, Kohl’s. I mean, we know what’s going on there. Like you don’t need to pay, right. Like you need some more winners. And, you know, thank God here for like Northwestern mutual.
Thad Davis: Mutuals a big supporter up there too.
Justin Nicols: You know, there’s not a ton of technology companies up here that are employing, you know, seven, eight hundred, two thousand people. That’s just not here, unfortunately.
Thad Davis: Correct. Yeah. What about, what about going backwards in time? What kind of advice, as we kind of wrap up the, wrap up the talk here, what kind of advice would you go back to that, Would you tell you, like, get better grades? Like, what’s that?
Justin Nicols: I think, I think that’s probably, that just creates optionality for you, right? I mean, as you know, that’s the best thing that you can provide. Anyone who’s like, “good grades, good university, better network, better options”, exactly. I mean, would that have created the grittiness of, you know, I didn’t grow up with a lot of money, primarily a mom who was the only breadwinner, primarily raised me, like, I hated being poor, not having money, like, so that created a certain level of grit in me, and there’s a lot of studies on that, that like, you need to be somewhat dumb or very risk averse to be an entrepreneur. And that normally comes from feeling not like having a lot of money and insecurity already because you need to be comfortable with insecurity. And so, you know, would that have changed a lot of who I was in terms of hunger? Because life likely would have gotten a little easier sooner if I went to a good school and maybe, you know, had a better network. I don’t know, but that’d certainly be number one. And the number two point to your point on like the region and what’s going on here. I think, for me, somebody who hopefully, you know, I’m not calling you in a week for a job is built to be an entrepreneur. Finding the right environment and company and mentor to discover that sooner is probably better. But as you know, that doesn’t mean start a company. That doesn’t mean try to go raise money. That means go learn from some people who are doing it very well. And that means unfortunately for this region, but it’s getting better, it’s like you need to go work at like a fast growth company then. Right? And really see these things and roll up your sleeves and do a lot of different jobs. and those are the things I probably would invest more and then just clearly more digital tools. Right? Like I, I can’t write, I mean, I can write an Excel, a little script there and there, but I can’t write a, R or Python. Like I should probably, you know, looking back, probably invest in that a little bit more.
Thad Davis: That should have been a yeah, that’s, I think I said that on another podcast. It should have been a computer scientist. I would be done. I’d be rich. And we would, we would not be on this podcast.
Justin Nicols: But would you have developed your business brain? You know what I mean? That’s the counterpoint here. Like exactly. And it’s like, yeah, you can get a good job as an engineer, but the really good engineers, the ones who have a business brain engine and engineering background. And that’s, that’s a different skill set.
Thad Davis: Yeah, that, that’s correct. Otherwise, you, you could be, you, you’d be a technical expert but not be able to create, like we talked about a minute ago. Perfect. Well, with that, I really, really appreciate the conversation today. It’s been good. We’ve, we’ve covered, we covered a lot of territory and I’m glad. always good to talk to you. I mean, things are always hopping with you, so it’s good that Sift and you continue to move things forward and I’m sure us and the, the listenership here will be very excited to see what unfolds over the next, uh, few years.
Justin Nicols: Thanks for having me, Thad. It’s great to be here and great to see you.
Thad Davis: All right, good seeing you.
Thank you for listening to this episode of our Perspectives podcast. If you are interested in participating in future podcasts or would like to learn more about Leerink Partners, please email us at info@leerink.com.
