Dr. Yusuf Sherwani on scaling Pelago, raising $150M, and the future of AI therapy

0:03Digital health is not for the faint-hearted. What seems like an overnight sort of success story is typically years in the making. Things are hot at times and colder at other times and I think you need to have this sort of mission-driven ideal at the core of what you're doing and this deep commitment to the long term to be able to wait it out.

0:31Yusuf Shawani, CEO and co-founder at Pelago. Thanks for joining me here at Startup Health. It's great to see you. Yeah, it's nice to be here again, Logan. It's been a few years since we last did this. Yeah, it has been really gratifying and really cool to watch Pelago's journey. It's been many years and yet folks might see you now and think you've been an overnight success, but as you and I both know, it represents years and years and years of hard work. So before we get into what you're building now and your view on the mental health landscape, just take me back, give me a bit of a flyover of the Pelago journey.

1:05How did you get to this desk in New York City where I find you today? It's definitely been a very long and meandering journey. And I think folks who have been building in digital health for a long time now just know the pains and the ebbs and the flows and the roller coaster that comes along with it. We started Pelago nine years ago. So this is pre-pandemic days. This is when virtual care was considered a bit of a novelty and people were questioning whether it would ever be acceptable or safe even to have a teleconsultation, particularly for something as stigmatized as a behavioral health condition.

1:40Then we went through the pandemic where something that was a nice to have became a necessity overnight and people got their first taste of it. And you went through this huge boom period of digital health where still there were naysayers. Like when you consider what we did in the substance use disorder world, there were questions as to whether or not controlled substances should be prescribed over telehealth or whether they would be abused. And the evidence ultimately determined that actually there wasn't any evidence of more abuse than you would expect at baseline. And then we sort of went through the bust years in between.

2:09Digital health is not for the faint hearted. What seems like an overnight sort of success story is typically years in the making. Things are hot at times and cold at other times. And I think you need to have this sort of mission driven ideal at the core of what you're doing and this deep commitment to the long term to be able to wait it out. But if you do, then it can have incredibly compounding returns. And that's just what we're starting to see in our space as well. Now, remind us just very quickly, what is the Pelago product today? Pelago today, and I can split it in some ways between Pelago 1.0 and Pelago 2.0, which is really how we think about this almost refounding of the company that we've been through over the last 12 months.

2:48But historically, we were a virtual clinic for substance use disorder, so a specialized provider really built around helping individuals with things like alcohol use disorders, opioid use disorders, stimulant, nicotine, across the entire spectrum of care, across multiple conditions, enroll in an evidence-based program that could really deliver a strong digital clinic user experience, a combination of a multidisciplinary care team and digital tools to help them in their journey. And it's a B2B2C model. So we partner with large self-insured employers and health plans across the nation, covering all 50 states, adults and teens.

3:24and the fundamental goal is how you get people to take the first step, how do you get them to great clinical outcomes, and then ultimately how do we bend the cost curve when it comes to really expensive and difficult to treat substance use disorders. We'll go 2.0. As we refer to it internally at the company has been what does it mean for us to have this mission in an AI era and how does behavioral health evolve given some of the challenges that our industry and sector interfaces. And by and large, I think what we demonstrate in 1.0 was by innovating in the underlying care delivery model, we were able to drop the cost of treatment for substance use disorders by 10x from a$30 ,000 traditional inpatient rehab to less than$3 ,000 per patient and get paid based on the value we were generating.

4:10Ultimately, we were able to turn what was traditionally a cost centre into an area of cost saving for our clients. And based on independent actual analysis really returns for five and a half dollars every dollar invested in the program and what we found was that that need of innovating the care delivery model and fundamentally delivering healthcare in a much more affordable way where we improve access while actually improving cost effectiveness wasn't just relegated to substance use disorder but by and large a challenge that behavioral health as a whole faced when you look at the fact that over the last five years it's increased four times faster than overall healthcare costs which on its own has been sort of double digit percentage increases for back to back years.

4:48And we saw this opportunity where sort of traditional mental health players were, if anything, adding costs to the system and resulting in sort of a shortage of providers being able to treat people, an opportunity to use newer technologies like AI, but also just the things that we are familiar with, like asynchronous care and telehealth, and really sort of re-imagine the care delivery model in this sort of AI era, which is really an evolution of the model from simply breaking down barriers to substance use care to expanding access to everybody's behavioral health care that actually works. Okay.

5:20So that's a good breakdown of what you've built. I want to talk for a bit to the side of our audience that are founders that are also building companies who would look up to you for sort of the success that you've achieved in terms of lives served, but also funds raised, growth that you've achieved. A quick internet search shows more than$150 million dollars raised. Is that true? That's right. Is that accurate? I see 11x revenue growth, 100 % client retention, 287 % year over year growth. Whether those are accurate up to today or not, that represents a trajectory that a lot of founders would kill for and that they would really look up to.

6:01So talk to me about some of the secrets to your success, some of the things that you did that allow you to raise more than$150 million, allowed you to have that kind of client retention? What were some of the choices that you've made along the way? Yeah, it's easy in retrospect to sort of take a look, but prospectively, it's often much more difficult to understand. I'd say like everything you described, client retention, funds raised, and sort of commercial success, these are like the way I look at them, they're sort of downstream consequences of ultimately creating value for our customers and making something that sort of people actually want to use.

6:41Our North Star and guiding principle has been we're a mission-driven company. You know, we fundamentally believe that everybody deserves access to behavioral health support that works, whether they have an addiction or a mental health condition. And if we're able to deliver it in a way that's ultimately productive and creates value for all the different stakeholders in healthcare, then it's just a question of execution and tactics and how you actually scale that out to the market. I'm going to push you on this a little bit, You raised$151 million and your answer is just make a great product and all the other things will follow.

7:16And we work with 650 companies in startup health and not everyone hits those points. And so I'm going to press you for maybe some actual decisions that you made that you think actually were inflection points. I know you have two other co-founders. I know you have physicians on your team. There's things about Pelago that are unique. I know you came from the UK to the United States. So what were some actual decisions that you made that you think made a meaningful difference in your trajectory? The part I would correct actually over there is I don't think it's a case that you can just build a product and then expect everything to fall in place after that.

7:55I think it takes real intentionality and experimentation and focus on distribution in order to get there. I think the real point that I would make is you've got to start on the basis of how do you create a product that fundamentally delivers value to customers, which has really been sort of our North Star from day one. And what I mean by that in practice is it sort of goes back to the saying of you want to be fixed on the big picture, the long term vision, but flexible on the details. We knew from day one what hasn't changed from a pedagogy perspective through the years is there is this huge unmet need in the addiction space.

8:32Ninety percent of people not getting access to care. And that actually costs a lot of money. And there isn't any good reason as to why that's the case. it requires real innovation in how you go about delivering care, which starts from a product perspective in creating something that is unique and differentiated. But once you've done that, I completely agree. You can't just rest on your laurels and expect successful follow. That is the classic first time I've found a mistake that we made in the early days. And from there, it really comes down to experimenting with distribution. And what we did was really tried to figure out who actually benefits when someone enrolls in the Pelago program.

9:08So we demonstrated very early on that we had great clinical success metrics. And then we tried sort of a bunch of micro pivots and failed experiments that didn't work. We started in the UK, like you said, we attempted to sell direct to consumer and we very quickly realized that the willingness to pay didn't actually match the amount of value that we were generating and the cost structure that we had. So DTC wouldn't work, people aren't used to paying sort of larger amounts of their healthcare. Then we tried selling to the national healthcare system in the UK where we actually had a lot of early success but we weren't able to get to the level of scale that we felt would have the biggest impact that we could possibly have then we started experimenting in the US market it was actually attempt number three that started to work when we were selling to employers so I think if there is a common thread or a learning over there it's being incredibly uncomfortable and being willing to like first of all recognizing that opportunity cost is a real cost and being willing to pivot and try out other things, even if, and the magical thing about product market fit is, you know, when you have product market fit.

10:08And if you're asking whether or not you have product market fit and you're seeing some very sort of a bait signs and you probably don't have product market fit. And we've never been afraid to kill the previous business in order to pivot to a better business, because that's what the company needs us to do. And when I talk about refounding, that's actually the type of DNA that I've been trying to bring back into the company at this point, because now we're a scaled digital health company with meaningful revenue scale. And if there's one thing that I'm always paranoid about, it's complacency and the unwillingness to reinvent ourselves based on where the industry is going.

10:39And using those examples of being willing to, you know, actually just pivot and genetics has been sort of a lesson that served us well. And I think needs to be relearned in order for it to serve us well in the future. Speaking of reinvention, I want to pivot a little bit towards this moment that we find ourselves in, in the field that you're working in, in mental health. And I want to understand where you think we're at in terms of the application of technology to help folks who are really dealing with mental health struggles and what you think is the right use case for AI in the mix right now.

11:16I think we're at this really interesting inflection point in behavioral health and healthcare as a whole, where we've never spent more money on healthcare and behavioral health, yet we've never received so little for the money that we actually spend. And that should surface all the fundamental questions that people have on like, where is that money going? Are we getting our money's worth for that investment in the first place? And take behavioral health as a microcosm of that. So we've never spent more in behavioral health, where we've seen 4x faster acceleration in overall reimbursement over the last four to five years, where inpatient and outpatient spend has increased more than 50%.

11:57More than half of people with a mental health illness aren't actually getting the level of support that they actually need. They're not seeing a provider or a clinician. You have employers and payers questioning affordability, but you still have huge unmet need in the market. So there is this fundamental question of the solution can't simply be training more providers and more therapists and more of the same. We've actually got to think differently. And I like to say in some ways that AI couldn't have come at a better time because it enables us to really think a bit more about what is the right cost structure for care delivery and what is the right relationship or sort of integration of human clinical support and sort of like technology-led AI support that's sort of cost-effective at scale.

12:39And if we take mental health is that easy example. We know that half of people in therapy today, they don't have an underlying depression, anxiety, or other mental health illness, yet half of people with mental health illness aren't actually in therapy, which is such a weird sort of dichotomy of situation. We built Sona. This is our new AI mental health specialist, which is designed to solve this care allocation and routing problem as a first line assessment, triage, and escalation layer that that delivers treatment and support to people at that low acuity end of the spectrum, where what they don't need is a therapist, but what they do need is a companion that's compassionate, able to listen to them and give them real evidence-based advice, but also recognize where there are clinical symptoms that might indicate a higher level of care necessary, and in a safe and effective manner, escalate them to a clinician that is able to see them.

13:32And I think in just like reimagining this care model, the whole goal is how do we get clinicians? It's not about substituting labor or sort of doing what a therapist would do, but how do we get clinicians working to the top of their license? Because we can't find clinicians fast enough for us to be able to meet demand. And if we can't do it, we suspect other people can't as well. And you've got to really think about how you get people working towards the top of their license and treating the people who are most unwell, who need to care the most. I think when people see Sona and they think about AI therapists or getting that counseling from an AI, a couple of things come to mind.

14:06One is a lot of people are already doing that with ChatGPT or with Claude. And so I'd like to hear really briefly where that difference is and what kind of information is specialized on a platform like Sona. And then on the other side is just a caution, any cautions that you have for folks who are leaning into AI as a first-line defense or even young people kind of pouring their hearts out to the machine. So maybe you could comment on those. Yeah, I mean, I'd like to say that ChadGPT is probably the largest provider of mental health care in the country by far and I don't think that's even a controversial statement to make.

14:41Now you can litigate the quality of that health care as you can with real clinicians as well but it's really interesting when you look at the stats and you find that half of people with a mental health illness have used AI support sort of in the last 12 months and I really do think that from a framing perspective you could look at this either from a glass half full or glass half empty perspective. On one hand, people are using it because they're getting real value out of it. The Harvard Business Review actually do this annual generative AI use case study where they look at the top 10 use cases and every single use case last year, bar the number one, changed in prioritization this year.

15:19And the one that remained consistent and in the number one spot was therapy and companionship. So that should probably tell us something about where we are as a society and the crisis that we're in generally. And I think the fact that people continue to use it mean that they are getting value from it in the first place, which to me presents an opportunity of what if you could deliver this in a much more clinically safe and accountable way, because there is a big difference between a great assistant, which is what ChatGPT is built to deliver, and a great therapist. And you actually see this in the underlying model spec where OpenAI have actually come out in the past in their model spec and said that their goal with ChatGPT isn't to fundamentally change people's mind it's to inform not influence but any therapist will tell you that actually you don't want to necessarily just reinforce people's existing thought patterns but the heart of cognitive behavioral therapy is changing the way people think so i do think that the opportunity is that people have demonstrated they want to have conversations in this form factor they like the fact it's on demand it's always available it's infinitely patient and it's willing to listen to you and it's non-judgmental people share more with sonar than they do with their therapist, even when they are seeing a therapist at the same time.

16:27So from our perspective, it's really how do we use this technology, refine it and make it much more purpose built, where there's always a clinician in the loop, there is always the ability and we're up front with Sona when people do their first intake, we say that, hey, a clinician may reach out to you if they feel it's necessary. And it doesn't actually stop them from sharing what they want to share. But there is a clinician in the loop, we're able to escalate things appropriately. But on top of that, for people who primarily want to use this AI modality, it's actually giving them clinically sound and safe advice that is based on real clinical practice.

16:58And that's the opportunity we have for much more purpose-built models, which is what we built with Sona. Okay. Well, that's almost our time. Quick closeout lightning round. One would be a word of advice to founders who are also building in the behavioral health space. And then finally, a word to our non-founder audience who might be interested in Sona and what you're building. Explain to them how they can get more involved in Pelago.

17:28So I'll start with number one, advice to someone building in the behavioral health space. I think there is naturally a lot of real excitement about the technology and the impact it will have. and you know I'm as AI pills as they come when it comes to sort of talking about the opportunity I think like you know we are sort of dealing with an incredibly vulnerable population where just safety and taking the right sort of evidence-based approach it's very paramount so you know my advice will be and this has always served us really well is you know go back to the clinical literature work with actual clinicians have clinicians in the loop from a product development process.

18:08That's something that's always sort of made us different and special. And I see no reason as to why that should be different for us versus anyone else. Continually validate things. We have clinicians annotating conversations that help to improve the underlying model and create sort of a flywheel of self-improvement. And then ultimately, where the rubber meets the road is in actual clinical studies that are peer reviewed where you're able to publish on the outcomes. You know, you can have the best engagement in the world and there ways of getting great engagement that don't ultimately move the needle on outcomes and outcomes is really what our industry needs so that would be my takeaway and then in terms of learning more about sonar about pelago it would probably be just visiting our website sonar.pelago health.com you can learn a bit more about how the model has been built and what's different about the approach we're taking versus what's been done before awesome youssef it's been really awesome to watch your nine-year journey and to see how you continue to reinvent what you're building, but also your approach to behavioral health in general.

19:07It's an exciting time and we'll be watching what you do over the next 12 months. So thanks for taking a few minutes with us. Amazing. Thanks, Lagan. All right. Take care.

19:19Yusuf Sherwani: And finally, an invitation to all of the extraordinary health transformers, Whether you are a startup, someone building the future of health, or whether you are an investor, funder, or buyer of the solutions out there in the marketplace, this is an invitation to you to join the Startup Health member community. Learn more at StartupHealth.com. Look forward to hopefully seeing you soon.