What happens when AI takes over healthcare admin?

About the show

Hosted by Nikola Mrkšić, Co-founder and CEO of PolyAI, the Deep Learning with PolyAI podcast is the window into AI for CX leaders. We cut through hype in customer experience, support, and contact center AI — helping decision-makers understand what really matters.

Summary

For years, AI in healthcare has been talked about in terms of clinical breakthroughs. But the transformation today may be happening somewhere less visible: in the admin work that surrounds care.
In this episode of Deep Learning with PolyAI, your host Nikola Mrkšić speaks with Alex Brown about where AI is actually making a difference in healthcare today. Rather than focusing on replacing clinicians, they zoom in on the systems that slow them down: documentation, scheduling, billing, and the countless tasks that pull time away from patient care.

Alex explains why healthcare fell behind other industries in automation, how administrative work steadily accumulated, and why AI suddenly feels like a step-change instead of another incremental tool. Together, they explore how technologies like ambient documentation and workflow automation are redefining productivity without trying to replace human judgment or empathy.

Tune in to hear:

Key Takeaways

Transcript

[00:00:41] Nikola Mrkšić: Hello, everyone. Welcome to another episode of deep learning with PolyAI. Today, I've got with me Alex Brown, who's our GM of health care. Alex, welcome to the podcast.
[00:00:51] Alexandra Brown: Thank you so much for having me.
[00:00:52] Nikola Mrkšić: Yep. So, uh, we didn't really have a big plan here other than not to turn the episode into just propaganda around health care being our fastest growing vertical. And the general kinda thing that I thought we could start with is just, like, you were an investor and then kind of, like, joined PolyAI to kinda help us grow out our health care vertical. What is working with, like, health care and technology and AI today? Like, what is what is the most exciting thing?
[00:01:16] Alexandra Brown: Yeah. And this is probably a little bit biased because now because now I'm at Poly. But also, when I was investing in health care, I looked a lot into AI and technology in the clinical space. So how do we kind of not replace doctors, but how do we kind of support the clinicians and the physicians as much as possible and takeovers what they're doing? And I think what I really came to notice was, like, it's a long journey. Doctors, physicians, they're very busy, and it's very hard for them to change on kind of the way that they're doing things. However, what we could actually help with is everything before and everything after that they need to do. All that layer of kind of administration, everything that's sucking up their time Yeah. That isn't actually treating or interacting with the patient.
[00:01:56] Nikola Mrkšić: Yeah. So, I mean, I think, like, one of the biggest uses of speech recognition technology historically was Nuance, Dragon, like, the software for kinda taking well, transcribing the doctor visit and then kinda, like, summarizing that, I guess, came a bit later. But that's, I guess, that post call wrap up.
[00:02:14] Alexandra Brown: Yeah. And I would say, like, if you look at AmbientScribe, they've they've transformed or the goal is that they're transforming how clinicians need to interact with patients. Right? Something's taking notes for you. We see it, like, on Zoom. You actually couldn't spend more time listening and less time typing notes. What I will say now is what we're seeing is, like, how then do you take that information that's transcribed and actually use it to then write a prescription or go downstream? Right? Do a bill. I think that is what we're seeing right now is the wave of, like, okay. We've got the information. How do we actually go take action on that information?
[00:02:47] Nikola Mrkšić: Okay. So, I mean, a lot of people when they come just, like, look at AI both taking over jobs and that whole fear well, the opportunity, I guess, on the other hand is that we all get infinitely scalable, high quality health care that is the best it could be for anyone. Right? And I guess that those appointments where, say, you're the doctor and I'm telling you why or whatever about my foot problems and, you know, it learns from that data and everyone else's data. In the end, it would probably be, like, the best doctor in the world because it's seen all that data. Right? Yep. But, um, would that kinda, like, issuing the prescription or that action that it's performing? Would that be, like, the training data effectively? Or is that how people are thinking about it?
[00:03:25] Alexandra Brown: Are people thinking about it that way? I think to your point, like, oh, we could replace the doctor completely. Actually, don't I don't think we as humans want that within health care. Right? So yes to that point. It would train. It would know what to prescribe. And potentially, when a doctor says, oh, this biologic is the best for you, it could give a nudge and say, hey. Just based on other patient data, these actual other biologics are also at record or could be provided or could have better effects. And so then you it kind of gives a recommendation engine to the doctor to improve maybe their practices.
[00:04:00] Nikola Mrkšić: Right. Right. Right. So then, basically, it would serve to accelerate them a bit. Yep. Give them other ideas, maybe improve their accuracy.
[00:04:08] Alexandra Brown: Mhmm.
[00:04:09] Nikola Mrkšić: But I guess, in where they believe the criminal responsibility to the doctor.
[00:04:13] Alexandra Brown: Yes. I mean, depends on how futuristic you look at this. No. No. No. Not at all. But, like, depending on how futuristic. The tech fiends will say, like, no. We can actually we're more accurate than doctors. Like, we can actually probably replace them completely. I think and we do a lot of this, right, with our dialogue designers or speech. Right? There's an empathy layer that we're not yet there with or that people expect that, like, they want from their doctor. That is not it doesn't have to do with accuracy. It doesn't have to do with efficiency. It has to do with warmth. And and I don't know at least what we've seen and what we're working on, we're not there yet. So we can improve it, but I wouldn't say we can replace it completely.
[00:04:53] Nikola Mrkšić: Yeah. I mean, I guess there's more than just the to play the tech fiend a bit. Right? But even if it was human and you knew that it wasn't human, you probably wouldn't be able to relate well, to feel the empathy because you would know that I mean, this gets philosophical. I don't know. People seem to be spending inordinate amounts of time talking to child GPT about their, you know, life problems, health problems as well. Right? I think you you told me that it's, like but definitely top three use case. Right?
[00:05:20] Alexandra Brown: Top three use cases.
[00:05:21] Nikola Mrkšić: Yeah. Yeah. Um, so I guess we'll evolve as humanity. I don't know.
[00:05:25] Alexandra Brown: We'll Which is also a little scary. Right? Because Yep. We know we know better than anyone guardrails around LLMs. Right? If your GPT is trained to confirm everything you're saying or bias against, right, and it's not medical device or able to give diagnosis, you enter a little bit of a scary territory when all of your pre thoughts. Right? You disclose only what you think your doctor should know. And, for example, if you don't wanna necessarily say that you are overweight or obese, and so you don't tell your GPT that. It has no visual way currently of seeing that. So it's a missed symptom or a missed indication on what it's actually going to recommend
[00:06:05] Nikola Mrkšić: to you. Okay.
[00:06:06] Alexandra Brown: Eventually, maybe we'll get rid of like, with wearables and connected technology, maybe it will have a full view of who you are as a human and your family history. But right now, I think what I've seen is it's missing a lot of inputs to be able to give a very accurate prediction.
[00:06:21] Nikola Mrkšić: Yeah. I mean, like, it's fascinating. Right? Because much like and probably more so than in other things we do. Like, your retail history with Office Depot is probably less of an interesting item than our medical histories of our families plus than, like, your, like, file. I mean, to get that best doctor, they would have to know everything about you. Right? And
[00:06:45] Alexandra Brown: And about your parents and about your grandparents. And I guess people underestimate how recent digitalization within health care has been. So it's like, okay. You might have an electronic health record if you're lucky enough. Right? I guess in The US, it is way more
[00:06:59] Nikola Mrkšić: I think that's interesting for for, kinda, like, the audience because this was new to me. And, I mean, having moved around a lot, you moved around a lot as well. It's kinda like you said to put a new doctor in your country. It's like blank canvas. It's like, who are you up? My name is Nicola. Right? And, like, a doctor looking at medical history, especially with, like, patients being so, like, transient and, like, in and out and, like, you know, we don't really have our own doctor. I mean, people do, I guess. But most people don't if they move their own a lot.
[00:07:27] Alexandra Brown: Less and less, I would say.
[00:07:29] Nikola Mrkšić: Yeah. So then, like, how in practice, like, how prevalent is it that a doctor has actually considered the full medical history?
[00:07:37] Alexandra Brown: I would say almost to none. When I for example, I moved over here. I had my doctor write, like, 10 bullet points on a piece of paper about, like, everything I've been through, and that's what I gave my doctor here. There's 50 things that were left out. So when they prescribed me all my new medications, because what's paid for here is not what's paid for at home
[00:07:58] Nikola Mrkšić: Yeah. Yeah.
[00:07:59] Alexandra Brown: They're missing a plethora of data, but they're doing the best that they can.
[00:08:01] Nikola Mrkšić: I think this is where, like, you know, you look at it. I mean, I I feel like every other serve is a complete, like, you know, hobbyist pharmacist and, like, you know, people over abuse antibiotics and all sorts of things just because, like, you have a cold. Might be pneumonia. Why don't you hit it? Right? And I I feel like they're like, in The UK, you look at it, and it's like there's a much higher threshold of, like, prescribing things. And it's very interesting to just kinda, like, think of that, like, future vector of a human being and, like, how you act on it. Because, I guess, different systems act on it very differently. Right? Mhmm. But, um, when you think of, like, just the promise of technology, and as an investor looking at this before, PolyAd, like, what are the most exciting things that are being transformed? Like, is it is it longevity? Is it kind of, like, preventative health care? What is it like?
[00:08:46] Alexandra Brown: I would say so I'm a biochemical engineer by background. And when I first started investing, I was so excited about, like, the possibility of, like, technology and AI. We were gonna find new proteins, find new drugs that were gonna increase us to 200. Like, it was so possible. And I think over time, what I realized, while still exciting, the things that actually take market right? Remember, I think you just quoted where it's like, oh, in in testing all the models do great, but where's the impact on GDP?
[00:09:15] Nikola Mrkšić: Oh, yeah. Yeah. Right? I mean, that that was, uh, Ilya's, uh, podcast where I think he goes very, like, a to b. Like, if the models are so good at evals, why is the economy not growing faster? It's like, okay. I think we gotta, like, delineate the factors in between. And I guess similarly here. Right?
[00:09:29] Alexandra Brown: Same in health care. Right? Like, the models are great in these test environments. Right? Like, we should have predicted so many new life saving drugs or completely almost replace the doctor, and yet here we are. Right? Cure cancer. And here we are not necessarily living longer or living better or healthier. Right? If anything, a lot of, like, chronic diseases are on the rise. So it's kind of like, what is the missing gap of bringing that AI into actual well, it depends on what you think the goal is. But if the goal is healthier living, sometimes where is gone short.
[00:10:01] Nikola Mrkšić: I see. I see. So then kinda, like, in terms of the investments being made, I guess, if you invest, you expect to get returns. Right? So if we follow the money, then, like, what is going on here? Like, what are we investing for? What is the, you know, kind of a cost function that we're optimizing for? Is it longevity? Like, when or is it just really completely chaotic?
[00:10:22] Alexandra Brown: Depends on the health system. I would say a lot of the large returns are actually with efficiency.
[00:10:27] Nikola Mrkšić: Okay.
[00:10:27] Alexandra Brown: Right? Like, how can we do more with less? And that's where kind of a lot of the big wins, at least in the the we'll call it the digital health side have been. Right? How can we serve more people with less humans?
[00:10:41] Nikola Mrkšić: Yep. Yep. Yep.
[00:10:42] Alexandra Brown: You have seen some great wins in the biotech space. Um, all the large pharma companies and medical device companies are heavily investing. But I would say yeah. If you're looking at digital health and services side, it's a lot in efficiency.
[00:10:56] Nikola Mrkšić: Yeah. I mean, that's what we deploy most of our systems. Right?
[00:11:52] Alexandra Brown: No. I agree with you. But I would say, well, you might in other verticals think like, oh, is that really innovation? Right? Within health care, it is. Right? Because it's a little bit we're a little bit behind in that kind of or we're a little bit slower to in financial services in open banking in Europe. Yeah. All your data is kind of available, and they've got great prediction on kind of, like, what you're gonna spend, target marketing, everything like that. If you don't even have an electronic health record, so it's in a paper file Yeah.
[00:12:20] Nikola Mrkšić: Yeah.
[00:12:20] Alexandra Brown: Yeah. Where like, banking is predictive modeling AI, the innovation that you're thinking about in other industries is light years away because nothing is yet digitized.
[00:12:30] Nikola Mrkšić: Yep. Yep.
[00:12:31] Alexandra Brown: Um, so we're probably only coming up to the cusp of the other vert uh, industries now.
[00:12:35] Nikola Mrkšić: Yeah. I mean, I've, uh, you know, I think that I have very naive, uh, understanding of this. And also color, I think, by, like, different countries and what we've seen through PolyAI there. And I mean, maybe talk a little bit about just, like, EHRs and, like, how is America so much more advanced given that you know, I think, to me, like, the high level numbers and you know much more about this, but what 17% of American GDP is spent some shape or form on, like, health care and r and d. I know part of that is, like, the fact that they're really fronting r and d development. Right? you think of, like, um, like, that part, like and then you look at just kind of, like, the fact that it's not, like, freely available and stuff, like, it is fascinating to me that public systems like Canada or The UK where, like, there are very kinda, like, loved health systems that are public and they're, like, the pride of the nation.
[00:13:49] Alexandra Brown: No. I mean, I would say there's no every country has tried a different model of health care in terms of the incentives and the payers. No no country is perfect. Right? Some can move faster. However, if you look at population health outcomes, right, necessarily that doesn't correlate. Why hasn't The UK or these kind of, like, government run succeeded in moving faster? I mean, Palantir, I think it was three, four years ago now, got contracted by the UK government to build this, like, federated lake of data, patient data that the they would then commercialize to pharma companies, and it would be a revenue stream for The UK.

[00:14:25] Nikola Mrkšić: Yep.

[00:14:26] Alexandra Brown: I don't think never happened. Right?

[00:14:28] Nikola Mrkšić: Is it in progress?

[00:14:30] Alexandra Brown: So they say.

[00:14:30] Nikola Mrkšić: Okay. I mean yeah.

[00:14:32] Alexandra Brown: Yeah. Right? And it's because the data is so parsed, and then everyone who you talk to actually knows what's in an electronic health record did a shit.

[00:14:41] Nikola Mrkšić: Really?

[00:14:42] Alexandra Brown: Yeah. You have doctors okay. Perfect example. We are talking to one of the largest RCM companies, revenue management companies. And the use case that they gave us is a doctor does shorthand notes in epic. Right? And then a nurse or practitioner calls up the patient if the notes are if all the blood tests came back okay. And they tell the patient, good news. Here are your test results. Everything is positive. But your cholesterol is within normal, but a little bit high, so do this nutritional you know, try to eat more chicken.

[00:15:12] Nikola Mrkšić: Yep. Yeah. Yeah.

[00:15:13] Alexandra Brown: The the thing. We're trying to read the clinician's notes. RAI is trying to read the clinician's notes. It's having a hard time because you have 10 letters. Right? And if RAI and we're spending a lot of time and effort on this,

[00:15:26] Nikola Mrkšić: it's having a hard time. Notes? Or These

[00:15:28] Alexandra Brown: are type notes. But, like

[00:15:29] Nikola Mrkšić: And they're still because I mean, the doctors, like, handwriting being really impossible to read. That's the oldest joke.

[00:15:34] Alexandra Brown: But they're like exactly. But think of that in electronic health system, unstructured data.

[00:15:38] Nikola Mrkšić: Okay. Okay. So it's just shorthands and things that people don't Just

[00:15:41] Alexandra Brown: shorthands. And so you can put it all in a data lake, but if it's not structured, I guess it's becoming better, but it's a little bit hard to do things with.

[00:15:48] Nikola Mrkšić: Sort of then you have that iterative thing where you have to, like, impose structure, wait ten, ten years to be followed by the data, you see what you missed, you fix it, you waited another ten years. Okay. Okay. Is there ways to drop by the way? I don't know. I'm taking, like, you know, you know, Apple Health was supposed to be any medical records or. Right? I mean, I I've not populated any fields. Maybe, like, height weight, but, like,

[00:16:13] Alexandra Brown: the people just announced it too, that they're gonna go into kind of the health side where it's gonna maybe eventually replace an EMR. Right? And they'll be able they'll they'll structure it that way. I think what you're saying

[00:16:25] Nikola Mrkšić: Of course. We know a lot of that. Sam wants to do commerce, and Daria is just gonna, you know, digitize help us. Us. Okay?

[00:16:33] Alexandra Brown: Okay. Yeah. K. But then you look at something like MonMed. We've had this discussion. Right? Like, they are end to end. So they do everything from kind of the intake to the revenue and the billing. Right? And what they're doing is they're trying to do it so that when their Ambient scribe doesn't just put a text in that copy pastes, it actually structures the data so that it's reusable later downstream. Right? So, yes, it's doable and it's a way to innovate, but you have to almost, like, own the entire patient journey Yep. Or or work with partners to really make sure that the data flows between.

[00:17:03] Nikola Mrkšić: That's so interesting. Okay. So then when you think of kind of, like, just AI's ability to help on that side and, again, my health care notation is out, but this would be, like, service kinda like

[00:17:13] Alexandra Brown: It would have to span, like, across service to the digital side to almost the administration of, like, therapies and things like that and connected devices.

[00:17:21] Nikola Mrkšić: Yeah. Yeah. Yeah.

[00:17:22] Alexandra Brown: So you almost need to connect, like, all three rounds of health care.

[00:17:25] Nikola Mrkšić: Yep. Which

[00:17:28] Alexandra Brown: is hard enough.

[00:17:29] Nikola Mrkšić: Yeah. Yeah. Yeah. Um, but, I mean, when you when you then think of kind of, like, just the ability of AI to help with different things, like, we forget about the analysis. Because in theory, it should be able to, like, look at that medical history. And I think one of the oldest examples of kind of, like, people saying that AI can be beneficial was radiologists. Right? Yep. And, um, I think one thing was, like, it is more accurate than humans. But in any case, it would like, a false positive is not such a big deal because you'll check anyway. And, like, well, I mean, if it detects something that is a false negative for the human, that'd save the life. Right? Yep. But I guess that's just another example of, like, more cost for better performance rather than actually, like, democratizing scaling it too. Although, I guess, you could use it for people who don't have that.

[00:18:16] Alexandra Brown: Was the investment case. So I've because

[00:18:18] Nikola Mrkšić: that's where people were looking

[00:18:19] Alexandra Brown: at. Right? And it's the the efficiency use case is why all the investors put so much money Yep. Into AI for radiology. It makes complete sense. You could do it faster. You're more accurate. Why not? The challenge is, and I'll speak for The UK specifically, is we didn't remove the humans for the process. They are double checking the AI Rad. So while it's more accurate, right, we are not seeing the cost savings that would kind of trigger mass distribution

[00:18:45] Nikola Mrkšić: I see.

[00:18:46] Alexandra Brown: And revenue.

[00:18:46] Nikola Mrkšić: I see. That we

[00:18:47] Alexandra Brown: wanted to see.

[00:18:48] Nikola Mrkšić: But if you kinda follow and I mean, it's another thing we touched upon earlier, but you look at just, like, investments and returns. I think what's fascinating right now I mean, okay. The current kinda, like, health fad is, like, what, GLP ones and the whole kind of, like that is, I guess, prevents the medicine. Right? If people regulate their weight, you'll have less health problems and long term less, I guess, what, cardiac arrest, strokes, and other things that,

[00:19:12] Alexandra Brown: like True. Yeah.

[00:19:14] Nikola Mrkšić: Would that be, like, a more kinda, like, traditional I mean, it's definitely paid off handsomely for the companies. Right? And is that, like, then a better example of where technology can help? Is that even technology? Or so it's definitely the AI model. Right? Yeah. But it's also I mean, it's fascinating how it's basically a recurring revenue model, where, basically, these things become software subscriptions that people buy to.

[00:19:36] Alexandra Brown: Well, that's the whole I think This

[00:19:37] Nikola Mrkšić: is the company.

[00:19:39] Alexandra Brown: The first, but, like, I'm thinking the blockbuster products, HUMIRA. Once every two weeks, you're injecting yourself. Mhmm. Couple thousand dollars. It's what you would call it software, they are a model, but for pharma. Yeah. So they love that is how we we do drugs. I guess, I personally think that's why selling cell and gene therapy was this breakthrough cure for cancer almost. Yeah. Right? It never took off to the scale that we needed it to because we couldn't figure out the pricing law. And it's difficult to deliver, and there's a whole other things. But I worked on a project specifically on how we were gonna price this for government.

[00:20:14] Nikola Mrkšić: What are we pricing?

[00:20:15] Alexandra Brown: We are pricing this drug, which is supposed

[00:20:17] Nikola Mrkšić: to be

[00:20:17] Alexandra Brown: kind of a more wonder drug, not an ARR model.

[00:20:20] Nikola Mrkšić: And it's wonder drug for doing what? Like, preventing Turing cancer. Cancer.

[00:20:24] Alexandra Brown: Let's say. Cell therapies. Yeah. Oncology.

[00:20:27] Nikola Mrkšić: So would this then not be, like, just a public good rather than a yeah. That stuff. Yeah.

[00:20:33] Alexandra Brown: How do you that's the problem. Right? Okay. I can't charge an ARR model, so I have to charge all the profit upfront. Yeah. Right? No government wants to pay for that.

[00:20:41] Nikola Mrkšić: Really?

[00:20:42] Alexandra Brown: They'll at least in The UK, it'll bankrupt them.

[00:20:44] Nikola Mrkšić: I guess because you you can't see why you're yeah. I see.

[00:20:47] Alexandra Brown: Over time. So then the pharma company says, okay. You can pay for it all up front. But if this person lives, right, you'll pay for it kind of over time. Right? Like, they'll be like milestones. If they live six months, you'll pay x. If they live twelve months, you'll pay y. Challenge there, which was really interesting, was government said, okay. Not the UK government. But what happened was then they knew it was gonna be so expensive. They were only giving it to the sickest where nothing else worked. Right? It wasn't tier one defense. And then farmer said, well, these people are dying anyhow because they're the sickest. They're not even making it to the sixth month. We're not making our money back.

[00:21:20] Nikola Mrkšić: Selling Gene.

[00:21:21] Alexandra Brown: Yeah. We can't pay it back.

[00:21:22] Nikola Mrkšić: Well, okay. But then I guess the place to a whole I mean, outcome based pricing is, you know, kinda like the team of the day. With

[00:21:30] Alexandra Brown: both voice AI.

[00:21:31] Nikola Mrkšić: Well, I mean, yeah, I don't know. I mean, like, he you know, it's, uh, there are many hot takes on it. One is like that. It doesn't really work because it's really hard to define upfront, you know, what that outcome is. The other one, the facts on the ground that I've seen play out many times when we did have outcome based pricing is that someone's paying you a mil. It scales. They're paying you 5. And then, you know, they do the math of wearable leaders. Like, they don't wanna pay 25 mil. So then I wanna do it on my own. And then maybe they can, maybe they can't. And it leads to, like, a really awkward situation where maybe they can't do it short term. Maybe you really have advanced technology, and you're the only one who can. At that point, you can kinda, like, hold them over a barrel if there is ROI, but eventually, it'll catch up. And then you've not really been a great partner in that you've charged them a lot. Now if you fronted the development of it, it's actually very similar to pharma. And I think in pharma and medicine, we at least understand that the cost upfront is such that it otherwise wouldn't be done. Whereas I think with technology that is a bit more mundane, let's say, customer service automation, you kinda go, like even though it is a very sophisticated piece of software models and all that, it's not all that different in terms of r and d. By the time it's commoditized, then it looks like you've just overcharged for well, that looks like trivial, picks and shovels. And it's

[00:22:49] Alexandra Brown: Well, then there's the whole if you take, like, software outcome based pricing, you went to health care. Right? Like, value based care. Right? So that's not fee for service. It's Kaiser does kind of some of these models. Right? Like, I will pay you for kind of the outcomes of how healthy this person is, readmission rates, things like that.

[00:23:07] Nikola Mrkšić: Okay.

[00:23:08] Alexandra Brown: Right? So then you're supposed to be looking over everything this patient does. And if they are healthier, you will make more money.

[00:23:16] Nikola Mrkšić: I see.

[00:23:16] Alexandra Brown: As a provider. There's pros and cons to both, I would say.

[00:23:20] Nikola Mrkšić: Good if it can be like, in theory, it sounds really good. Right?

[00:23:23] Alexandra Brown: But then you need a little bit maybe like a Kaiser, an integrated care network where the payer also has the provider. Right? You need that you need to be able to prove both ends. Right? Yep. Yep. This person is healthier because of what we're doing.

[00:23:34] Nikola Mrkšić: Rather than just

[00:23:35] Alexandra Brown: Here's the baseline. Right?

[00:23:36] Nikola Mrkšić: I've got good genes.

[00:23:37] Alexandra Brown: Yeah. Because then you just curate, oh, I'm just gonna serve these patients that have good genes. Right?

[00:23:41] Nikola Mrkšić: Which global health care systems are, like, the best at, like, both having the control of that and then kind of, like, also the ability to curate the data and iterate on this? Is it is it in The US mostly? Or but when we look at longevity, I guess America is pretty good.

[00:23:55] Alexandra Brown: I guess the challenge with The US, and I'm not, I would say, like, a US population health expert or anything, is you would need to subsegment it. Right? Who has access to the Kaiser Permanente's of the Right? And then once you go with that group, then you can see kind of the trends.

[00:24:09] Nikola Mrkšić: I'm surprised you don't see ads of, like, work with us because our people live longer than, like, that network's people because that would be a real, like, outcome based brain.

[00:24:19] Alexandra Brown: That's how they get the best doctors, right, and physicians. So it's an interesting every country, I think, has a health care incentive program that's a little bit different.

[00:24:27] Nikola Mrkšić: Because I think the okay. I mean, UK, Canada have, like, public Yep. Health care. That's like Canada has only public. Right?

[00:24:34] Alexandra Brown: Yes. They're transitioning. I believe there's now. I'm originally from Quebec, and there's a there's a private wave coming up.

[00:24:39] Nikola Mrkšić: Is Quebec more hardcore on it being that, like, non private or in line with the rest of Canada or less?

[00:24:46] Alexandra Brown: French Canadians, I think we're probably moving a little bit faster in the private space.

[00:24:49] Nikola Mrkšić: Okay. Okay.

[00:24:50] Alexandra Brown: Be on. Yeah.

[00:24:51] Nikola Mrkšić: Okay. Interesting.

[00:24:51] Alexandra Brown: Left. Yeah.

[00:24:52] Nikola Mrkšić: Okay. And then, um, when we look at just kinda, like, other uses of AI in health care, like, what else is interesting?

[00:25:00] Alexandra Brown: To me, like, all the connected predictive things that are going on where I could get a notification, but I'm also I I like to do health. Right? To do something before my GP has to see me.

[00:25:10] Nikola Mrkšić: Uh-huh.

[00:25:11] Alexandra Brown: That to me is interesting.

[00:25:13] Nikola Mrkšić: So as in, like, what, like a blood test before you or no. That's mundane. You mean more like

[00:25:17] Alexandra Brown: No. Even a blood test or perfect example is, like, I used to always have to go in, let's say, a mold test. Right? Now they send it to your house. You hook it up to your iPhone. You're able to scan yourself. You send that in. You get a text back next day. You're good to go, or please come in.

[00:25:32] Nikola Mrkšić: Oh.

[00:25:32] Alexandra Brown: Like, to me, that

[00:25:33] Nikola Mrkšić: Well, that's from us health care more available. Okay.

[00:25:36] Alexandra Brown: I think that's personal just because I have a passion for it. I think democratized health care is most exciting to me. Although if you ask a health like, a a health care purist, they would see a little bit more around the innovation of, like, drug discovery and longevity is more interesting to them. Right?

[00:25:54] Nikola Mrkšić: Yeah. But I guess it goes hand in hand just in terms of, like, if you can at mass scale, even, like, from, like, Fitbits and, like, Apple watches gather, like, these signals. You should, in theory, be able to go, like, you seem fine. Don't come in. Something's off. Something's changed with you. Come in. How do you feel about these, like, companies like NECCO and, like, others where it's, like, full body scans and, you know, all the way to, like, Brian Johnston and his, like, you know, measuring everything. Everything

[00:26:18] Alexandra Brown: Tracking.

[00:26:19] Nikola Mrkšić: Like, yeah.

[00:26:19] Alexandra Brown: I think I actually as a healthcare person, I love it. Right? I do think that the follow through it's great to go get an echo scan. What are you doing with that information? Right? Are you going back into the NHS? Right? Are you waiting another two years to get the elective surgery? So I think once you have that information that they call it, like, the referral gap or whatever, how are you actually getting the treatment that you need is where kind of that crack still falls through.

[00:26:45] Nikola Mrkšić: I see.

[00:26:46] Alexandra Brown: Yeah. You know? If we're talking about voice AI and we're talking about poly, what's exciting me there is we are really good at doing, as I say, like, that administration layer. We are relieving health care staff of kind of that burden so that they can actually have that empathetic conversation. But if you're looking at the frontier of voice AI, people are diagnosing or working on, they're not yet there, diagnosing mental health, insomnia. They're trying to do diabetes through voice.

[00:27:10] Nikola Mrkšić: Through voice.

[00:27:10] Alexandra Brown: Right? Yeah.

[00:27:11] Nikola Mrkšić: How does that manifest? Like, you said They're

[00:27:14] Alexandra Brown: using biomarkers in your in your voice. So we're talking to companies right now.

[00:27:17] Nikola Mrkšić: Do you need to have, like, the voice, like, sample before and after? Or Correct.

[00:27:22] Alexandra Brown: Okay. Yeah. So they track it over time. We're talking to a company right right now that's thinking about building on poly to do right? Where they would they would use our platform to track those samples

[00:27:32] Nikola Mrkšić: Oh, wow.

[00:27:33] Alexandra Brown: And then be able to spit out

[00:27:35] Nikola Mrkšić: a diagnosis. Wow. So the voice does con contain, like okay. Well That's the theory.

[00:27:40] Alexandra Brown: Okay. Okay. Yeah.

[00:27:41] Nikola Mrkšić: That's really cool. Which could be a

[00:27:43] Alexandra Brown: a next frontier. Right? You're calling in to make an appointment, and at the same time, you detect that.

[00:27:48] Nikola Mrkšić: Got it. Well, it's it's upsell opportunities. Did you know you you

[00:27:52] Alexandra Brown: need this Okay. Capitalization. I like it.

[00:27:54] Nikola Mrkšić: Okay. Okay. Okay. No. That that's really fascinating. Um, Yeah. I mean, like, just you know, I hear everything you're saying about, like, the need for, like, digital transformation so, like, you get more benefits. But I've been really incredibly positively surprised by the uptake in health care. Right? Because, you know, I think a year and a half ago, health care wasn't in our, like, top four verticals, and that's number one. Right? It went from, like, one to 10,000,000 in a year. And I'm like, the appetite and the innovativeness. And, like, I don't know. There's something I mean, with Gen AI, I feel like what what's kinda, like, flipped around is that enterprises are more aggressive than, you know, people who maybe have a lower risk factor and are able to there's a lot of top down pushing. But I think when you look at verticals, like, health care has just been so forward. And we're not the only ones. There's so many people in different pockets of this growing to, like, really, really substantial figures. Yeah. Why do you think health care has, like, the actual, like, number one adopter?

[00:28:49] Alexandra Brown: Yeah. I would say if you ask me probably, like, seven years ago, like, health care and AI, I was I was like Oh, don't even try. Right? Right? It's so Like like, it's a it's a mess. But I think at the time of this disaster, we were looking at the health care and clinical space. Right? Right? And it took so long. And I think Might have started to be described, to be honest. Right? That was probably the first time when we were like, oh, it actually doesn't need to be AR radiology making a diagnostic.

[00:29:15] Nikola Mrkšić: Yeah. So that's your

[00:29:16] Alexandra Brown: But, like, it's not that serious, but it's transformational. I think people underestimated how much of health care's employee time is spent on admin. Right? And that's why it's like they saw Ambiance Web, and they were like, oh my god. What else can we do? And that's why they've all turned around. I think we get interest from, like, one practice, one physician running their own clinic to thousand clinics. Right? They hate the admin side of things, and that's what we're helping to alleviate. And I think in other industries, you've made incremental changes to alleviate admin. Right? In health care, you really haven't. So it's kind of making that massive leap where they're super So

[00:29:57] Nikola Mrkšić: it's catching up in terms of that Gotcha. Okay. Yeah. And because of the regulation, the whole, like Yeah. Paperwork has always been important and mandated even though the notes the notes may be comprehensible, but they have to be left there. Right?

[00:30:08] Alexandra Brown: Correct. Especially with how the the incentives work around payers and fee for service. If you're gonna change to value based care, you really need to document everything.

[00:30:15] Nikola Mrkšić: Yep. Yep. Yep. Yep. Do you think, like, um, I mean, like, I guess, like, one theory and one of our investors, uh, Vinod Khosla, I mean, he's always talked about just kinda, like, the abundance of health care you could provide, for example, in India, just the scale in places where you're just not providing adequate health care right now. I mean, the other place I think of naturally is, like, Serbia and Eastern Europe more broadly where, you know, the demographic pyramid is completely screwed. And then on top of that, all qualified laborers leaving disproportionately, leaving such a shortage of doctors relative to to the population. Do you think some of these places will then end up having to kinda, like no. This will be more progressive, but through sheer need, they might have to kinda, like, say, hey. Tier one triage is AI first. And then in theory, like, while it may have some catastrophic effects early on, it shouldn't theory lead to kinda look a better health care model, more data.

[00:31:07] Alexandra Brown: Well, so I mean, it's I have so many theories on it.

[00:31:09] Nikola Mrkšić: And unethical experiment that might just be forced upon. Yeah. I don't know

[00:31:13] Alexandra Brown: how dangerous and, uh, it's a little controversial. I don't know how dangerous and unethical it is anymore. So there's a PhD. He's fantastic, and all he does is health care AI. He's at King's. He tried to raise money in The UK to do exactly that, an AI first clinic. Think of it like a smart clinic.

[00:31:27] Nikola Mrkšić: Yeah.

[00:31:28] Alexandra Brown: Um, I unfortunately couldn't invest, but he couldn't get the traction here, and he's going to do it in in a developing country. Okay. Because the regulation is lost, and he wants to prove it out and then bring it back here. And he's designed this head to toe where it's all either AI virtual. There's, like, not a human at the clinic, and he's predicting I don't know how far he is. I should probably touch base with him. But, like, that the quality of care that they're gonna get might be better than, like, what we're seeing here.

[00:31:58] Nikola Mrkšić: I mean, a 100%. Yeah. I mean, I you know, just kinda, like, when you when you think of all the kinda, like, DIY health care and everything, I remember, uh, my father had COVID. He had refused to go into a hospital because he was like, I won't make it out alive. Very rational. But I remember my cousin who's a doctor and I, like, just from, like, the general notes that were global, looked at, like, corticosteroid, like, amounts, everything, etcetera. And it was very possible and, you know, like, it was fine. Again, not proved that we did a good job. He might have been fine anyway.

[00:32:30] Alexandra Brown: I would like to think it's a pretty accepted truth that staying out of the hospital is actually healthier if you can. It's why the UK government is trying to push everyone out. US is trying to care at home. US, I would say too, if they can find a way to pay for it. Right? Because sometimes they get paid on beds, but care at home is better as long as you can make a diagnosis without seeing the person, which then some people have questions about.

[00:32:53] Nikola Mrkšić: Oh, wow. That's really fascinating. It does make sense anecdotally, but, again, I have no data. Okay? Okay.

[00:32:59] Alexandra Brown: Doctors visiting you at home. I mean, Ceracare, which does exactly that. Like, nurses, I think they'll eventually IPO. They're kind of one of the European's big success stories.

[00:33:09] Nikola Mrkšić: Okay. And that's what it was. Visits at home by

[00:33:11] Alexandra Brown: Visits at home.

[00:33:12] Nikola Mrkšić: By quality.

[00:33:13] Alexandra Brown: Nursing and yeah. Like, um, medical staff.

[00:33:16] Nikola Mrkšić: Okay. Mhmm. So I guess that's an efficiency play, but still, like, the quality comes from not being in hospital. Correct.

[00:33:22] Alexandra Brown: And more people choosing to not go to hospitals if they don't have to.

[00:33:26] Nikola Mrkšić: And then I guess because of that, reducing the burden of them getting medical help for something that they might have ignored. Yeah. Okay. Well okay. So there is, like, a lot of potential here that is happening right now.

[00:33:37] Alexandra Brown: Yes.

[00:33:37] Nikola Mrkšić: Yes. Okay.

[00:33:38] Alexandra Brown: And I would say in the I definitely do think in developing countries. I remember when I was studying health care, there was a teacher that was kind of talking about here, we do cataract surgery one by one. It's like a thirty minute whatever process. You would never. I think it was in India where there was an innovative doctor that said, line everyone up in one room, 15 people. Right? Just make sure it's sterile, and we'll do them all at once. I'll move from bed to bed.

[00:34:03] Nikola Mrkšić: Look at Chez Grandmaster playing, like

[00:34:04] Alexandra Brown: No problem. 15 people in and out. Right? And it was super efficient. Everyone was fine. And then those 15 people started a support group because I think there's, like, um, post surgery adherence that you have to adhere to. Right? And they were bonded, and they showed that, like, actually, not physio, but, like, they were keeping to a better schedule than those people that just got operating.

[00:34:24] Nikola Mrkšić: Really? Because it

[00:34:25] Alexandra Brown: because it built a sense of community.

[00:34:26] Nikola Mrkšić: That is fantastic.

[00:34:27] Alexandra Brown: So, like, to me, that's innovation too.

[00:34:28] Nikola Mrkšić: Oh, no. 100%.

[00:34:30] Alexandra Brown: Yeah. But it's so against what we would probably do in the West. Never don't even wanna share a hospital room. You're not gonna wanna share surgery.

[00:34:37] Nikola Mrkšić: No. That makes sense. I mean, like, you know, when you were saying that, like, we want the doctor on the empathetic side, I think that we have had a few examples of better engagement with some mental health use cases with, I think, people disclosing either financial difficulty or things like STDs, right, where, like, you don't necessarily want to tell another human being because you're ashamed. I think those were, like, the two examples of where AI actually had an edge. Mhmm. Because, well, it's not human. Right? But, um, yeah. Okay. Awesome. Well, look, this has been really, really fascinating, and I hope the audience will will enjoy learning a lot of this too. I definitely learned a lot. Uh, but thank you for joining me. Please like, share, subscribe, and, uh, we'll see you in the next one.

[00:35:22] Alexandra Brown: Thanks for having me.