Virgílio 'V' Bento on building and investing in the current AI moment

Uh, how's your week going? It's been a week. Yes, a long week, I think, for everyone. Yeah. Uh, is it specifically because of tariffs or AI news or what's driving that? Yes, tariffs, the impact. But some other stuff, it's like it's been fine. It's been fine.

Uh, can you just give us a a quick a quick inter introduction for the listeners who might not be familiar with Sword Health? uh kind of big quick backstory on the company and what you guys do. Yeah. So let me start with the problem.

So healthcare is quite special because when you compare healthcare with other industries right like the consumer electronics industry right in the last 40 years you saw penetration of technology in those industries that went like this right so massive penetration of technology. What happened?

Sorry my alarm just what was that? It's my alarm. Sorry. Oh, no worries. One second. We have some technical difficulties, folks. Okay, I'm back. Sorry. And I was saying that in the last 40 years there was a massive penetration of technology in the consumer electronics industry, right? What happened to cost?

Massive decrease. Right? So, a true vision that 40 years ago or 30 years ago would cost $3,500 now cost $500. Mhm. Much cheaper, much better display, much better features. So we use technology to produce better goods. Yep. When healthare last 40 years, you also saw massive penetration of technology. Mhm.

What happened to costs? Contrary to the other industries, massively increase where actually we use technology to make healthcare more expensive. Mhm. And the reason why is because the way we've been using technology in other industries is to shift part of the labor from the human to the machine.

And with that, we made the produ production of goods much higher quality, much more efficient, and much more accessible. Mhm. In healthcare, we've been using technology to double down on this 100% labor intensive model where before for you to have an appointment with a physician, you'll go to an hospital or to a clinic.

Right now, you can use a technology to do that through a video call, but you are still fully dependent on the human on the other side. Right now, has always been for you to access healthcare for 30 minutes. you need 30 minutes of this highly specialized, scarce and non-scalable human resource which is the clinician.

So what we believe that in order for you to deliver the future of the healthcare world, what you need to do is basically develop technologies and AI that will shift part of the labor from the human to the machine and with that you remove barriers in terms of access.

And so what we are doing at SW is really shifting health care from that human first model to that AI first model and we starting with the biggest problem which is how you deliver care how we recover patients back to a full life and so we started with physical pain we expanding to pelvic health and now we are expanding into other verices of care really changing how people access care with AI.

Can you talk about uh Jevans paradox was in the news recently um and I'm sure you had a lot of thoughts just because you know theoretically as a cost of health care declines we're going to want more of it there's a lot of care that doesn't happen for example uh in physiootherapy if it costs you know a dollar a day I'm sure people would you know be much more eager to use it but when it's $200 a session or even more than that um so I'm I'm curious uh feels like an you know uh I'm sure you've seen this playing out and I'm sure you weren't um weren't worried when Deepseek came out and and you know uh token costs came crashing down you probably were like okay this is great we're just going to use a lot more of this yeah and look lack of access to high quality high-intensity conservative care non-invasive care doesn't decrease costs because when you have pain instead of you are not able to go three times per week to pity cleaning for three months and that would solve a problem right but what you do next is you go and try to find a silver bullet in the form of surgeries right and that's what really skyrockets prices just with physical pain in the US we are spending 560 billion with a billion dollars per year right and the big bulk of that is surgeries which should be replaced with much better outcome for patients right and so when you use AI to make the tradition traditional model which should be the solution very easily accessible as accessible as running water then you allow patients to get access to high quality care and that's how and then you really decrease costs because the problem in healthcare right now and by the way this is in the US but this is all over the world right the post the national health service in the UK is suffering the same thing which is you have very high high costs and very very low quality and access to care so in the traditional equation in in healthcare.

If you want to increase quality of care, you need to hire more people. If you hire more people, you increase costs. But you cannot increase costs because costs are already prohibitive. So what you try to do, you try to reduce cost to remove folks from the equation.

What you do, you reduce quality and you basically create massive challenge in terms of access.

The only way to really break this paradox is by using AI doing part of the job of the human and then having the clinician in the loop highly scalable in order to be able to bridge the gap between the massive demand and the low level of clinical supply that we have. Yeah.

Uh how has your AI thesis around healthcare changed since founding the company? if it's even changed at all.

Is are the thought is what you're seeing today and the sort of roadmap today, you know, been what you expected or has your has any sort of like uh new technology that's been introduced in the last few years sort of changed? Weren't weren't you just talking about um AI therapy breakthroughs recently?

There was a yeah there was some news I think earlier this week that basically showed that AI therapy in the form of people just talking with LLMs like actually delivers fantastic results and it's basically free right and it's like this incredible it seems like a fantastic add-on if you have a scaled business you've been in the business for 10 years you don't have to start from scratch on the customer development journey uh so yeah talk to us about that we we we actually in that regard we did an instinct experiment because we have Phoenix which is our AI system based on the sessions that the patients doing at home preparing the messages for our clinicians to send to patients right and so based on they analyze the session of the patient and based on that they can tell you hey John I saw that you did your session perfectly in the end you were a little bit exhausted so I decreased the session a little bit for you to be able to do it without a problem right is the work that our PTS our clinician usual usually do right and now of course we have LLM looking at the data and preparing those messages, right?

But we were looking at the messages and we had this question mark that do the messages feel like it's an AI creating the messages or do they feel human? Right? So what we did was a blind test where we had 50 messages from that were written by our clinicians and 50 messages that were written by Phoenix, right?

And what we wanted and then we asked our team to evaluate which messages were from the clinician and which messages were from from Phoenix, right? And what we wanted was a 50% randomness where you cannot distinguish which messages are from which uh Phoenix or clinician. Oh wow. Like pass the turning test basically.

Yeah, exactly. What we got was very very surprising because the messages from the clinician were mainly labelled as coming from the AI and the messages from the AI were mainly labelled as coming from the human.

That's bizarre but I understand and that was and then we we went a little bit in more detail and what we found was that look since the clinicians are always like going from one patient to the other and they are always in this almost state of burnout across healthcare the messages are very dry and very concise y right where the LLM the messages are very warm and then also they have this long memory where they can pick up things from two weeks ago that you said right they can pick things about stuff that you said that I want to recover because I want to play with my kids and go on hikes and you said that during the ongoing form and of course human ne never remembers that right and so it's funny because the AI and LLMs make the work of the human of our clinicians more human than before and that's super funny.

Can you talk a little bit about uh AI image generation? It seems pretty far out for what you're doing.

But at the same time, I'm just thinking like if I need to show someone an image of, you know, hey, your knee, I have a diagram of your knee and this is where, you know, the the the the physical damage is, this is what we're going to be rehabbing.

Uh, I could imagine even just doing the basic like studio Gibli style transfer could just make that whole experience feel a lot less medicinal and a lot more enjoyable and just novel. Um but have you even started playing with those tools or is AI image generation uh still pretty far out on the road map? No.

So our focus is really on um when the member so what we want to do is replicate the experience that you have in the clinic with a clinician replicate at home with finish right and so we have the feedback component which is we analyze we observe what you are doing and we provide feedback right then the corrective feedback that's where we are experimenting with that type of layer of imagery layer because then you can basically say hey do the movement like this, right?

And or do the movement like that.

Um one one area where we are using that and we are exploring uh how we can do that in more detail with AI is we have a solution focused not on physical pain but on pelvic health and pelvic health is basically things like urinary incontinence after child birth which is a massive problem in female population right which basically is you it's implied for we have an interventional sensor where you can basically have to train your pelvic floor muscle no one knows what is the pelvic floor muscle and so using imagery to say hey you have this thing which should contact like this and using AI to make that animation much more lively it's how we are experimenting but that's the thing it's it's like a can be an explosion of possibilities with AI because you are using AI voice agents right we are using that to enroll patients right we are using AI to identify that member that in six months going to have a surgery So we can intervene now a little bit like minority report to before that person go to to the orthopedic surgeon and get convinced that they need surgery we can act right now we are using AI to quantify the motion of the patient.

So it's really can be an explosion where everywhere I look I see an application of AI with clear benefits. That's great. Yeah.

How do you uh you guys have been very successful very quickly uh in in relative well I'm just saying overnight success yeah overnight success in a sense from a pure revenue ramp from a revenue ramp standpoint and what usually when a company ramps revenue really quickly other people go hey that's a good idea we should do that too how ambitious are you guys as a company is is the sword health for x sword uh or or you know or is it just about running down these opportunities that that you're currently tackling?

So basically yes uh our view is really translated in shifting healthare from human first II first.

So any every single area of health care which is still delivered in a 100% labor human labor intensive way it's the target for us to reinvent right one one area where you apply that is well mental health care right like the way you address mental health right now is talk therapy where you should basically solution to mental health is talk to that human once per week or once every two weeks right that's 100% human level intensive that's an area for us intervene, right?

Um and so it's really about all luckily for us in terms of addressable market is is the fact that like pretty much everything in healthcare it's 100% human level intensive and so we have a very aggressive market um road map in terms of expanding and replicating what we did with physical pain what we did with pelvic health into other verticals of care because the thing is when you really nail product market fit healthcare it's ma it's massive in terms of expansion because you don't saturate right we the pelvic health solution that I was telling you about Bloom we launched that solution in 2022 everyone thought it was a niche solution we did in that year $500,000 of revenue we did last year $25 million this year we're going to do $50 million wow right and everyone thought I I still remember discussing this solution with my board and my board saying yeah don't focus on that because that's too niche right and it's like it's exploding And so healthare the good thing is when you get the product market in healthare you have untapped growth potential because the market is just crazy big.

Well congratulations. I mean that's all amazing news. Uh what a fantastic industry to be in at time. We spend three hours sitting every day. We need uh we need some PT Phoenix for the guy. Yeah. Okay. Perfect. Perfect. Well, thank you so much for coming on the show. We'll have to have you back when there's more news.

We'll talk soon. Thank you guys. Thanks a lot. Yeah, the the growth there is just shocking. Yeah, and it's like such an underrated company because uh they're out in Portugal and they do have an office in New York City, but um it's just one of those uh like under the radar companies in my opinion. Totally.

Well, we got another guest coming on the show, but about wander first. Find your happy place. Find your happy place. Find your happy place. Book a wanderer with inspiring views, hotel grade amenities, dream events, top tier cleaning, and 247 concier service. It's a vacation home, but better. Go to wander. comtvpn.

Fantastic. Uh, we got Avlock coming in. Uh, I'm sure everybody All right, there he is. Avllock, how you doing? What's going on? What's going on, guys? How you