Johannes Schildt & Claes Ruth, CEO and CFO of Kry, on the AI future of telehealth

But the beautiful thing about it is, there's no upper limit. You could build a massive company in this space—it's 10, 11% of GDP in all European markets—stable, in a way because demand will never go away and you have really large payers.

How do you think about the role of public payers, such as the NHS and county councils, in the evolution of telehealth services? Why does the NHS (and other national health services) want to work with businesses like Kry? Are Kry the only ones working with the NHS, for instance, in this capacity?

Johannes: The public healthcare system is based on public/private partnerships, so a lot of the public healthcare is delivered by private companies.

Previously, there were no frameworks that allowed for delivery, and delivery models were not in the same room as the patient.

Now policymakers and governments want it. They're asking us, "What should we do? What policies should we change for you to start to invest in our country?"

Before, we were knocking on a lot of people's doors, and people were shutting them in our faces. Now, it's the opposite. One of the reasons why may be that they’ve finally started to realize that their existing way of structuring healthcare just will not work.

Governments are really bad with innovation in general. You have a lot of examples of that across the world, where considerable capital is pouring in but it never works. The model of having private companies delivering public services is not uncommon in Europe but usually it’s not one big company providing all of it.

We see that you have a bit of that dynamic in one way, because as all of the U.K. contracts that we are running, it's not about exclusivity. But if they have a partner that works well, why would they look for someone else? However, it's structured slightly differently in the different markets. If you asked me a few years ago, I would probably say that policy and regulations, that's one of our main challenges and risks. It’s less so now.

Over the last 12 months, you have reduced EBITDA burn by more than 70% and are projecting profitability across all geographies by 2024. What did you do to achieve this significant reduction? What are the key pieces of your plan to reach profitability?

Johannes: Our plan obviously was always to be a profitable company, but we’ve made a dramatic shift towards profitability and increased our profitability a lot during the last 12 or 18 months.

The context around it has obviously been that the market has started to value different profiles of companies, from valuing growth to valuing profitability. We had high investment levels there for a while because you have to play the game—there were smaller companies that were popping up, raising large seed rounds, and we had to protect our position by investing quite a lot. That’s less important now—because we don't see that competition anymore.

What we're seeing right now is that demand is being increasingly aggregated to us. Both public and private partners are looking for stable partners, and that's us. We see a lot of the larger private payers, because even if our main customer is the public system, we are payer-agnostic. We don't care who is paying for it, so we’re also partnering with health plans. Some of those smaller companies have been partnering with some of them too, but they just haven't reached the scale where they can operate that at a profit.

What we’re going to see is that a lot of the competition will disappear and demand will continue to rise, and we are one of the very few that can continue to deliver those kinds of services, which is a really good scenario for us.

From an operational perspective, it's been tough for us during the last year to do significant reductions and continue to operate, grow, and do new things. But we have managed to do that very well and squeezed a lot of efficiencies. Before, where capital was less expensive, we could also push some of those efficiencies to the future. Now, we have way more operational focus and are just improving efficiency across our teams and services. That’s been the main focus during the last year.

Claes: What's noteworthy there is also that we managed to cut costs dramatically while growing organically. Not only did we grow the number of patients, but we also improved service efficiency at the same time. It's really tricky from a CFO perspective to see companies that have succeeded with what we have done in improving all metrics because typically, you succeed in either growth or cost-cutting. But we've actually managed to do both at the same time, and improve metrics across the board. That's positioned us, as you said, 70% closer to profitability. Still some way to go, but we're moving steadily in the right direction.

Given the complexity of those relationships that you have with governments on the operations side, do those function as a moat for Kry? What is the Kry “secret sauce” to forming those kinds of relationships with governments?

Johannes: It's a trust business. But the moat is never one thing. We need to be good at different things simultaneously, and we create a lot of value when we're combining strong tech and product talent with strong medical talent. There are companies that are doing only one or the other but it was clear to us very early on that you can't drive significant change by just doing the technology layer. You also have to go in and operate.

Our secret sauce, in a way, is that we are very strong on the tech and product sides—on the healthcare operational and the commercial sides—since we address both, the private and public payers. On the public side, and with all payers, it's a trust business. Now, when we've reached significant size, we have 25% of the Swedish population signed up for the service. We deliver more healthcare than the largest European hospitals, and by all standards, we're a huge healthcare provider, and we’re doing that. We are the only ones in the U.K., on the digital side, that have been rated outstanding by the Care Quality Commission, so we have raised the bar for medical quality and quality control.

We have built that trust. That takes time and it was tricky in the beginning because no one knew who we were. We didn't have a well established brand, the founder and I were not from the healthcare sector, and we didn't have the top medical talent that we now have on board. There's multiple things that we need to be good at simultaneously—in healthcare and what we do—and that's the complex part. But it's also the fun part and a big part of the moat.

We are a local healthcare provider in all of the different markets. The rules and regulations are slightly different, but combining that quality assurance with pure healthcare delivery side entailing a strong product and technology, is tricky. Given the significant local integrations to prescription databases, we're also handling a lot of patient information. It's a lot of things, but that said, it's also one of the reasons why there's no one else that has scaled outside of the home market.

Now we're also starting to be really good at finding those synergies, because even though we have the payer infrastructure and a legacy system infrastructure, there are obvious differences between the United Kingdom and Sweden and France. However, there are many things that are similar. So we're really starting to be good at finding those synergies across the markets.

Also we’re not being forced to do it. Earlier, we were running local teams in all of the different markets, so one of the reasons why we've been able to reduce healthcare costs quite significantly is that we removed that lens. Now, we run full functional teams across all of the markets, which works, but a lot of that entails finding those synergies in the models, in the technology, and on the operational side so you can replicate it across markets, even though languages, rules, regulations are slightly different.

In the last few years, Kry has opened 28 new physical locations in Sweden, and four in Norway and two in France. Can you talk a little about how you’re approaching bridging the physical and digital, especially given the challenge of doing anything capital-intensive in this current economic climate? How does bridging the gap between physical and digital make care more efficient by keeping patients ‘in-network’?

Johannes: There are two things.

One is the purely regulatory side where in many markets, the reimbursement models are usually tied to physical locations. That’s stupid, but that's the way it is structured. Another big part of it is that from a product and consumer perspective, you don't want to build a digital service that is a siloed system where you can help people with everything that can be done digitally but if they need stitches they need to go to someone else. While you can start there, that's not great.

What we have been doing now, successfully, is knitting together the digital pathway with the physical pathways because obviously not all healthcare will be replaced by digital. There will still be a part of it that would need to be physical. In Sweden specifically, and if you look into Europe in general, you can simplify by saying that you have two models in healthcare—the fee for service models and the capitated models, the subscription models.

In Sweden specifically, the majority of the healthcare expenditure in primary healthcare is on the capitated side, and that's a lot of good business for us because it's stable and a subscription-based model that’s tied to clinics. That’s one of the reasons that we need to run that physical aid of infrastructure. Also, by having both, obviously you could serve way more patients per physical doctor or physical clinic, but yes, it's both product and product-proposition that we need, together. Then, it's a regulatory thing.

Over time, we don't run all of that ourselves, obviously, so we don't really care if it's us or someone else. It's been very beneficial for us to also do that at scale in the Nordics so we are within a controlled environment, with our own assets. You can then start to link it all together.

In France, it's unlikely that we will fill the need to go heavily into the physical element. That will be there but we won't have to operate it ourselves. We can do that through partnerships and they can send you somewhere else for blood samples, for example. We did that before we were running healthcare clinics ourselves, and we still do. We have 600 locations across Sweden alone where you can go and take blood samples. We don't operate those, but we partner with them, and you can take your tests there. The data though is fed back to us, and then, we act on it.

So yes, you need to be good at linking those two worlds together. You think, in the very long-term—eventually, yes—that whole infrastructure will be completely obsolete. But that will take decades.

Can you talk a bit about the decision to employ your own in-house doctors and healthcare workers? Why did you go with that over the more ‘asset-light’ model of facilitating connections between existing healthcare workers and patients?

Johannes: It's not that easy to drive significant change by just giving tools and technology. We do that—60% of all GP practices in the U.K. use a subset of our tools—but usually what happens if you just give the existing system access to technology is, they will say, "Oh! Thank you for this piece of technology" and they will put it in the drawer. Or, they will continue to operate the same way as the day before, but just move the visit itself to video. They will still be like, "Oh! I run my clinic, I open at eight, I close at five, I'm closed on weekends" and that will not significantly improve the patient experience.

We have a solution where you have 24/7 more or less access and you get it within 20 minutes. That would never be possible if you just added this as a layer on top of the existing system. Then, is that a business to sell software to healthcare workers and hospitals and clinicians? Yes, obviously it is. Doctolib is doing it. You can build a good company by doing that, but if you look into healthcare expenditure, healthcare IT is a tiny, tiny, tiny part of healthcare. We are using our own technology to deliver healthcare at scale, with mainly our own resources, but we are always very open to let others borrow and use our technology.

If you look into our clinician base right now, a lot of them are employed by us, and some of them are contractors. Some of them work full-time for us, some do it in the combination with doing something else like working at another clinic that is owned by someone else or being retired, or even being on parental leave. It's also a beautiful thing that we are good at unlocking underused talent, where you have a lot of retired doctors that are usually super senior because they work their whole lives, but they might not have the energy to run around in the emergency room. So, they happily do a few hours a week digitally with us. That's how we do it.

Can you share some details about how Kry plans to integrate AI into its services? Are there specific areas such as diagnostics, patient monitoring, or healthcare management where you think AI can bear the most fruit? In the broader sense, how do you see AI transforming the telehealth and digital health industry over the next decade?

Johannes: We’re doing really cool things. It's very clear to us that healthcare is one of the areas that would benefit the most out of large language models and AI in general. It's also very clear that the service sector will benefit a lot from this, because you can significantly increase your modules.

We are partnering very closely with OpenAI and Microsoft, and we already have some great things in production.

From a regulatory perspective, you will still need access to contracts, patients, and doctors to do it. But the companies that will benefit the most are those like ourselves—have significant scale already and access to patients and clinicians. No one else has that at the same scale as we have, and the others that do, they just run a physical hospital and they can't implement it.

So what are we doing right now?

We are more focused on not replacing the clinician, but giving them superpowers, removing administrative work, and helping them be more precise in diagnostics.

To give you a few examples, we use OpenAI's Whisper to listen to calls. It's more precise on diagnostics and helps set secondary diagnostics which is a revenue driver for us because if we are better at coding, we get paid better. We don't have that in production yet, but the next step, I think, would be to start doing more of the analysis before the consultation with AI (artificial intelligence).

From a patient perspective, you start by interacting with our service. We ask the relevant questions that the doctor would ask, and then, we don't let them make a decision and present it to the patient. We field that back to the doctor, so, one of our doctors would reach out to you. The doctor will then get, "Here's the symptoms, here's the diagnosis codes, here’s the recommended treatment. Do you agree with this?" Then, over time, in the majority of cases, doctors will be, "Yes, this is right. Thank you." Those models have read all of the medical journals ever created.

In a not too distant future, we will be able to significantly reduce the administrative work for the clinician, or completely remove that. We will be able to speed up and enhance efficiencies on the clinician's side. We are in the business of selling very expensive conversations, and that's something that the large language models are very good at.

Claes: We're on the brink of a massive change in the way we produce, administer, and quality check healthcare. So even though—regardless who's right, Johannes or I, on exactly when or how doctors would be redundant—we're on the brink of a revolution in terms of capabilities, we are one of the front-runners in applying the new technology.

Johannes: What’s cool is that the core technology is ultra complex, but it's not that hard to apply it. The really heavy lifting is a lot about the piping around it, and we have that. When GPT-4 was released, you literally had tens of thousands of startups that were completely obsolete, because there's this whole market of small layers based on large language models—just useless.

There are companies doing exactly what we do. We're listening in, helping you listen in to calls, and helping with diagnostics. We would never buy that. We’d just build it. The companies that will benefit most are companies—and it is a bit unfortunate, because we just cement the larger companies' position—that already have traction in patients, or customers where you can apply it. That's doing the analysis robots where you're prompted so it can run the diagnostics and be very precise about it. Anyone with a decent AI team could do it now, but then what? What would you use it for? You don't have the patients, you don't have the contracts, you don't have the payers. It's useless.

However, it's super cool. Also, it's changing so dramatically. Our approach has always been, "Someone will build the fundamental models. We will not do it. Why would we invest in it?" There's a lot of companies that have invested an awful lot of money trying to build their own models. Yes, eight months ago, the best standard for us would have been to train the models ourselves with a lot of our own data. You don't need to train that anymore if it already knows the local guidelines and the ICD-10 code system. That is also really cool because it's moving so fricking quickly. It's cool because you will have a lot of companies be like, "Yeah, we're going to run it in our customer support, yada yada."

We can actually apply it in the core model and in what we do. What are we doing? We are delivering healthcare. So yes, we can apply it there and not in the services around it. We're like, "Oh! We can increase efficiency in our customer support a bit, or we could generate illustrations quicker." We can actually apply it in the dead center of what we do, which is cool.

Do you see it potentially changing the business model in terms of how people pay for healthcare? How Kry gets paid? We spoke to companies about how substituting SaaS with something built on an LLM could affect how they have to think about pricing. Is that at all a consideration?

Johannes: I think the general trend is towards what you in the U.S call value-based models.

Eventually, we will end up there in most markets, but the shift is not happening quickly, and it’s in different phases in different markets.

In France, it's a fee-for-service model. My belief is that it will continue to be so for quite some time, but with the exploration of AI (artificial intelligence), I think we’re definitely heading for capitation value-based models—half of our revenue is now on a capitated value-based model.

The capitated model and value-based model that we operate in Sweden is slightly different from a U.S. system where you just take full risks. If our patient gets severe cancer or whatever, we don't have the financial risk, so we manage the risk on the primary healthcare side, but not on the secondary and acute side. But yes, long-term, I think that most of our revenue would be on a subscription based model, where it's more of an all-you-can-eat model on the patient side.

If everything goes right for Kry in five years, what does it become? How is the world changed as a result?

Johannes: Our ambition has always been to build the largest healthcare company in Europe. Period. Not the largest digital healthcare company, but the largest healthcare company.

It's very clear to us that the largest healthcare companies in the world, in Europe, will be digital and tech at the core, and it will be very hard or impossible for the old legacy providers to make that shift. We know them well. They're spending a lot of money on innovation every year but they're not going anywhere.

I think it's the same thing with a lot of industries. It's very hard to change. It's hard to become a tech company. Tech in healthcare has always been viewed as something like, you tender for it, you buy it, or you have someone in the basement in the hospital that buys a system. It’s not in the incumbents’ DNA.

This transcript is for information purposes only and does not constitute advice of any type or trade recommendation and should not form the basis of any investment decision. Sacra accepts no liability for the transcript or for any errors, omissions or inaccuracies in respect of it. The views of the experts expressed in the transcript are those of the experts and they are not endorsed by, nor do they represent the opinion of Sacra. Sacra reserves all copyright, intellectual property rights in the transcript. Any modification, copying, displaying, distributing, transmitting, publishing, licensing, creating derivative works from, or selling any transcript is strictly prohibited.