📣 Digital Health Download: October 2026

0:02Steve Kraus: Welcome to the Heart of Healthcare podcast. I'm Hallie Teko.

0:07Michael Esquivel: I'm Michael Esquivel. And I'm Steve Kraus. And every Monday, we bring you the latest in healthcare innovation as we sit down with entrepreneurs and industry experts. So buckle up and join us as we figure out how to improve healthcare for all.

0:32Michael Esquivel: Hello, listeners, and welcome back to the Heart of Healthcare podcast. I am your co-host, Steve Krause, and today I'm joined by my friend, Michael Esquival. Michael, happy fall.

0:42Steve Kraus: Happy fall, Steve. Great to be doing this episode. So, Steve, before we kick off today, I just wanted to give a shout out to you and your partners at Bessemer on closing two new funds, raising a total of$5.75 billion. Super excited with what you all are going to do with it, And just a big congrats to you and your partners.

1:01Michael Esquivel: Oh, thanks, my friend. Excited to be able to invest more in great health AI companies. So if entrepreneurs are listening, we're open for business. Can't wait to hear what you're working on. Hey, Michael, let's get started. We have so much news, lots of news hot off the press this month. I'm going to start with something that literally got released this week. And so excited to talk about it. As you know, we have an investment in a company called The Bridge, which is one of the leading scribe companies. This story is a story for the medical legal nerds out there. Michael, at least I'm glad we have you as a co-host this week because you're definitely a legal nerd.

1:34Michael Esquivel: And so excited to get your perspective on this story. This case, it's a case that's called Raphael versus Montai. It's in Washington State. And it basically has to do with the fact that a patient had a telehealth visit with a doctor. And that doctor, they didn't use a bridge, but they used Microsoft DAX, which is another ambient scribe, which records the patient visit, the conversation that was captured and generates a draft summary that's then put into electronic medical record. Obviously, the physician can edit that before it goes into the note. And in this particular visit, the physician actually tried to basically talk through her unsuccessful attempt to find the patient a rheumatologist for a follow-up.

2:15Michael Esquivel: And then at the end, actually, the physician said that she just was gonna withdraw from being her doctor. Don't know why the reason that was. But the legal case hinges on the fact that under a Washington Uniform Healthcare Information Act, it gives the patient a right to examine and copy their health information, what's in their EMR. We all have access to our own medical records. That's law and to our medical information belongs to the patient. But it carved out information that was compiled solely for things like litigation, quality assurance, peer review, or administrative purposes. And the patient in this case said, hey, this is my information.

2:52Michael Esquivel: I should have access to it. And the courts now just in one state ruled actually know it's protected. The enterprise is protected from sharing that information because it met one of these carve outs, which were like we said, litigation, quality assurance, administrative purposes. In this case, they said administrative purposes. On the one hand, Michael, this is probably a good thing for enterprises from a liability perspective and may actually lead them to be more comfortable with adopting AI scribes because they're protected. And again, it's not final law, But that's one way to look at it. But on the other hand, I mean, I'm a big patient's right fan and the rights to their medical information.

3:29Michael Esquivel: And this seems to fly in the face of it. I'm curious, like, where do you come out on this one? Because this is like really interesting, like law happening in the sort of AI world as we speak right now. Yeah.

3:40Steve Kraus: You know, what's really interesting, Steve, I want to just double click on something you said just a moment ago, which I think is really critical for our listeners. Look, this is just a single state trial court decision. So let's keep this in perspective. It's not binding precedent outside of the state of Washington. And it's another example of the emerging unsettled area of AI law and AI governance. It's yet another example of where the regulatory framework is trailing the technology, innovation, and adoption pace. So it is interesting from that perspective. You know, I thought about that a lot, too, Steve, and, you know, not a perfect example for us lawyers, but, you know, there is a qualification in the law for work product that lawyers prepare.

4:24Steve Kraus: You know, it's in context of a litigation, so, you know, not quite analogous. But the idea, it sounds like, from a conceptual framework was, look, these transcriptions, these AI scribe transcriptions are more like dictation or handwritten notes. And so, you know, the conventional use of those dictation or handwritten notes is that they're temporary aids that the physician uses to building her final chart. And so the analogy is an interesting one. It's a lot harder because handwritten notes, you know, pure, you know, scribing transcription, it seems like it goes further than just a handwritten note or a dictation.

5:05Steve Kraus: But I can see the analogy and why the court sort of ruled the way it did.

5:09Michael Esquivel: And your argument is like it's an input, not the actual output. Exactly.

5:14Steve Kraus: And I think I think that's where the court concluded. Like, it's an input. It's it's it's in the process of producing that medical record. Right. That record of care is, I think, how the statute defines it. So it doesn't reflect, I guess, in the end, the clinical judgment, because it is a transcription. If it had the clinical judgment, and that's where the final written note is, that final clinical note, is what is the official medical record.

5:44Michael Esquivel: Yeah, our portfolio company, Abridge, actually, the reason this sort of aims me a little bit, this ruling, and I understand some of the rationale, is that our company, Abridge, everyone knows now because it sells to health systems. But originally, SHIB started that company focused on consumers and particularly folks who are diagnosed with cancer. And, you know, you go in with a loved one or yourself and you're diagnosed with cancer, like the medical terminology is flowing at you fast and furious, right? And it's really hard. Obviously, emotionally, you're basically been hit with a bat. And then you have to absorb all this terminology and figure out what to do with it.

6:17Michael Esquivel: So that's actually where a bridge started first is like transcribing some of those tough conversations and turning that language into understandable lexicon for patients and then helping them determine what the right next steps are. So I'd hate for a ruling like this to take away that opportunity for patients to really be involved in the conversation and then understand the conversation after they leave the doctor's office. because it's so complex, a lot of these clinical conversations, that I think patients having that output is really important for them in terms of digesting and absorbing what was actually said in the doctor's office and the next steps, right, and follow-up.

6:50Steve Kraus: For sure. You know, for me, I look the counter to that, and I completely agree, by the way, because it is confusing for those of us that are not physicians, that don't understand the lexicon, as you noted well. But I'm hopeful that these transcription technologies will make the written clinical record clearer and better, right? I mean, maybe it's not going to be the transcription per se, but hopefully they're more robust and it's a little more consumer facing in terms of the ability to digest what's actually written in there in the final work product that becomes that clinical record. So.

7:25Michael Esquivel: Well, interesting story to follow because this will probably, there'll be follow on cases. You're the legal expert, but this certainly I assume will be appealed and follow on cases, I would presume. Yeah.

7:36Steve Kraus: You know, it's going to be interesting because one of the tougher aspects for our industry is when you have a bunch of patchwork rules across different jurisdictions. And the last thing you want is 50 different set of rules that you have to try to figure out how to comply with all of them. So it'll be really, really fascinating to see, Steve, does this set a precedent for other jurisdictions, other states to follow? And what does this mean? But to your point, I think the story's not done in Washington state alone. So we'll see where this nets out. Well, Steve, so speaking of hot off the press, Let me turn to another topic.

8:09Steve Kraus: Our listeners probably heard or read that on September 29th, Aura announced that it was postponing its IPO just days before it was expected to public debut. It had been expected to start trading the week of the 28th. They were going to price on Tuesday the 29th and then presumably start trading on the 30th or even today, which when we're recording this podcast October 1st. It was really interesting in the timing. It launched its roadshow on September 21st. As many probably read or saw, they were targeting an offering of 50 million shares at somewhere between a price range of$40 to$44 a share. At the top of that range, Aura would have raised approximately$2.2 billion at a fully diluted valuation of$15.6 billion.

9:00Steve Kraus: So, yeah, the company noted in that announcement that it was postponing that despite strong demand due to uncertainty in the IPO market, it was postponing. So it was an interesting statement. As of today, we don't know how long Aura's IPO will be postponed or when the company plans to return to the market. but it certainly raises a broader discussion about how companies and their legal and banking advisors think about timing of IPOs. What was really interesting about this one for me, Steve, is how the offering was structured. As many likely saw, most of the IPO was going to be secondary, right?

9:37Steve Kraus: Many of the major investors were going to be participating in the IPO and selling their positions. The company was raising a smaller primary component, and that was designed largely to satisfy RSU tax obligations that would come to pass post-IPO. And those were tax payments related to those employee stock awards. So it was a little bit uniquely structured in that sense. What's surprising to me is you don't normally see deals pulled at this late stage in the process, You know, presumably, as the company went through the roadshow process, it was getting feedback on valuation along the way. And so I thought it was it was interesting.

10:23Steve Kraus: I guess the silver lining, there's so many different ways we could talk about this. But a silver lining for Aura here, of course, is it doesn't need to go public, Steve. As you know, it's been profitable since 2024. Last year, there was 60.8 million dollars of net profit on one point two billion dollars of revenue. So the good news is they have the flexibility to decide when and if and how to approach the, you know, coming, trying to come back to market at some later point. What did you think, Steve?

10:50Michael Esquivel: So, Michael, I, you know, I think it's clear that, as you said, that Aura is a great company, right, and can stand on its own, even if it doesn't get public. But I do think we have this overhang of, like, the public markets wondering whether the digital health market is real. And I think Aura is grouped into that, even though it's a little bit of a different business model and form factor. And so one of the things, even though they are profitable and they don't need primary capital in the raise, like probably lots of shareholders going and selling in secondary into the IPO probably spooked the public market investors who haven't made money in our sector and see people running for the hills, or at least that's the perception.

11:26Michael Esquivel: I think the other thing is obviously the book wasn't massively oversubscribed, right? As you know, Forex is not that healthy of a book. And the worry there is just that a lot of the book is made up of hedge fund. you know, hedge funds that are going to trade in that very quickly and cause volatility in the stock. So I think that's an issue. And then of course, finally, is I think, and you're the Mr. Wearables, you can comment on this in that, you know, there's lots of competition in this market. And so like how unique is Ori, even though it's a really cool form factor, like the output is kind of the same as Whoop.

11:57Michael Esquivel: And we even saw this week, Apple make a real big play with their Apple device in longevity and in connecting with labs. And so starting to like combine behind what looks like a function with like a Aura and a Whoop. And so it's just a very, it's a very competitive market. So I think that's the overhang that Aura faced. I'm not an expert, but I'm, the competition was the biggest question I heard from my friends in the public fund market is like, yeah, Aura is really cool, great business, but like, how does it differentiate itself long-term?

12:26Steve Kraus: Yeah. And of course, with Apple making its announcement just recently, Steve, I think you're right. Just for perspective on the oversubscription of the book, Steve, to your point, I've seen you sort of think about, and there's no science to this, but from experience, you see 10x or more sort of a healthy oversubscription. 20x, you're in like the euphoria stage. And Figma, for example, reportedly was approaching 40x oversubscription. So I think that probably all all factored into this. But, you know, it leads to a broader question, what's going on in the IPO markets generally, right? And uncertainty that Tom mentioned, Tom Hale, the CEO of Aura mentioned in the press release is certainly there.

13:13Steve Kraus: I mean, so we've had, you know, the Fed hiking interest rates for the first time in three years, right? We've seen mortgage rates now topping 7%. Average gas prices, you know, have exceeded in states like California,$6 a gallon. And of course, most recently, what we're seeing from some of the leaders in AI raising fears and alarm bells about what's going on with that technology broadly. So I think that all played into not only Aura, but a lot of what we're seeing in the IPO market. So let me just share a few stats that I think were interesting as I was sort of surveying what is going on in the IPO market.

13:47Steve Kraus: So it's actually, putting aside that Aura may have been the only digital health listing in this year, it's actually been pretty healthy. It's picked up. I mean, the SEC tracked 208 IPOs in the first half of this year. That's up 16 % from last year. And the amount of money raised jumped from$28 billion to a staggering$138 billion. Now, I will note that in those stats, the SEC does count SPACs and closed-end funds that are registered under the Investment Company Act. But what was interesting is I dug a little deeper. traditional corporate IPOs in that group actually declined from 111 in the first half of 2025 to 88 in the first half of this year.

14:30Steve Kraus: So I guess on balance, I wouldn't say the IPO market is completely back, but it's probably more accurate to say that capital is flowing again, but the reopening has been a bit uneven. So digital health is interesting. And as you noted, shout out to Hinge Health. I mean, they're up. They IPO'd in May of 25, Steve. You were involved in that at$32 a share and closed yesterday for us, September 29th, I should say, was$95.21. And I think it's trading above that recently. So it's more than a 3x return right now for that IPO in what, 16 months, 17 months, something like that? That's what we need.

15:11Michael Esquivel: We need public fund buyers to make money in our sector, right? That's the point.

15:17Steve Kraus: Well, you know, it would be remiss not to just point out, though, that the M &A markets are booming, too. Just quickly, Steve, you know, we've seen, you know, very recently AMD announced the purchase of Fei-Fei Li's World Labs for$8.2 billion. We saw SpaceX's purchase of Cursor earlier this year, NVIDIA and Hugging Face, and then Stripe's recent acquisition of OpenRouter. So it feels like the channels are there, The opportunity is there. And I just hope that for our sector in health tech, that we do get a few of these IPOs out in 2027.

15:50Michael Esquivel: Let's hope. Well, I got my next story is about what I think is going to create the IPOs of the future, which is another health AI story, Michael. And, you know, it's interesting. A lot of the talk recently is, you know, the promise to AI is that it will, you know, increase access, you know, probably improve quality of care as a result, but hopefully ultimately decreased cost, which has always been like the, what's weighed down our industry, it's just how expensive it is. Now, to date, Michael, as you know, and we've talked about this on the pod, has been inflationary, not deflationary in terms of healthcare spend, mostly on the admin stack, particularly in the RCM use cases.

16:25Michael Esquivel: But the real hope I think is that, you know, AI helps our industry become deflationary, that total cost of care goes down, especially when this AI, these technologies are applied to clinical, the clinical realm and clinical use cases, right? And potentially even when it leads consumers to get more involved in their own care. But there's a story I read this week, actually it was in New England Journal of Medicine. It's in a section called Voices. So it's more about doctors sharing their experiences. And this one made me think whether AI actually is gonna deflationary on the clinical front. And the story is the following.

16:59Michael Esquivel: It was a cardiologist at Cedars-Sinai who described her experience receiving a message by a patient from a patient portal where the patient said, quote, please arrange for a coronary CT angiogram to assess my risk of coronary artery disease. Like it was almost like an inbound. Now this was a patient of the doctor's existing patient, but it was like, hey, I want this doc, go get this for me. And the doctor described themselves in this missive as being sort of irritated by the tone of the question. And then, you know, almost like it was a restaurant order, you know, like it was an entree they were ordering, right, the patient.

17:34Michael Esquivel: And then the doctor looked into whether it was medically necessary and said that actually the coronary CT angiogram is only recommended for individuals with chest pain, which this patient didn't report. In addition, not recommended to patients over 65 years of age because of inaccurate interpretation limited by kind of some of the classifications of the actual diagnostic. And so the doctor said, therefore, I must decline your request to have this test ordered. and the patient ended up ghosting this doctor. And the point of this piece that the doctor wrote is like, what do I do? Like, I lost this patient who is a customer, who is revenue, but on the other hand, like what they were asking for, which clearly came from an AI prompt, whether it be chat GPT or Anthropic or what have you, like they clearly did the research.

18:24Michael Esquivel: The doctor's like, we're gonna see more and more of these. And like, what do we do? It's fine if the patient's paying out of pocket, but they were asking for this to be in reverse as well. and it doesn't make sense from a medical health economics perspective. So like it was the first time where I said, I love consumers being involved in their healthcare and I love the fact that AI is gonna help this, but at the same time, doctors are gonna be put in these really awkward positions where it's their word or it's their expertise versus, you know, AIs. And so curious reaction to this story that I told you about, Michael.

18:56Steve Kraus: Yeah, you know, it's fascinating, Steve. I wonder, and maybe this is just human nature, If the directive had been posed as a question instead of a directive, would that have gotten a different reaction? And sort of saying, please order, saying, you know, I've been doing some research and I think it might be relevant to me. What do you think about my, you know, about us ordering this test? And it would have been, I'm just curious if human nature would have reacted a little differently. Look, you and I have talked about this on the pod. Like, I think patients having more information, the ability of some of these incredible AI tools to educate you, as we talked about in the last story where, you know, you've got an ability to understand the interaction with doctors.

19:41Steve Kraus: You know, these AI models are really educating us in a way that historically wasn't possible. So there's this tension, and I wonder if a little bit of the reaction was really in response to, like you said, pulling up and saying, okay, can I have a cheeseburger, fries, and a drink, please? Maybe that was some of it. But it is going to be a really interesting ethical question because we know the Hippocratic Oath, and we know their obligation, and we do know that sometimes if you run enough tests, you're going to find things that may seem problematic, but in reality, they're not. And so where is that fine line, Steve?

20:18Steve Kraus: It's going to be really interesting to see as we continue to empower the patient consumer about her health, how they're going to engage with doctors, and in turn, how doctors are going to respond to those queries.

20:31Michael Esquivel: Yeah. Well, we'll see. Anyways, what do we got next, Michael? Yeah, okay.

20:35Steve Kraus: So sort of dovetailing on your theme about inflationary versus deflationary effects, some of our listeners may have seen that CMS administrator Dr. Oz acknowledged at at a recent summit that AI will increase healthcare spending, that inflationary effect, before it reduces it. His reasoning was that AI is making existing billing infrastructure much more effective. His phrase was that it will turbocharge billing. We're already seeing evidence of this. You might have seen, Steve, the Blue Cross Blue Shield Association study recently that estimated that increases in billing associated with these tools added nearly$1 billion in healthcare costs from 2023 to 2025.

21:19Steve Kraus: Now, this$1 billion was healthcare spend where there was no change in care delivered. So we've seen other larger numbers, but this is where you held the steady state of no change in delivery, where were the increase in cost was a billion dollars. Providers, of course, argue that AI is helping them capture diagnoses and services they previously failed to code. And insurers, in turn, argue some of that represents just higher reimbursement without more care actually being delivered. So it feels like we're entering this era of the battle of the bots. Providers using AI to maximize billing coding, upcoding maybe here, and payers using AI to deny more claims in response.

21:59Steve Kraus: So it's an interesting battle of the bots in that sense. Or is it, Steve, really, this question for you, is it really correcting historical undercoding? And what do you think? Because Dr. Oz has a few more thoughts about how we might reach a deflationary moment. But for now, what's your take on this?

22:19Michael Esquivel: Yeah, well, I think the thing that you highlighted, which is really interesting, I didn't actually know this. I saw the headline about the BCBS study, which was a billion dollars in healthcare spend, but I didn't see the part that it represented a higher reimbursement without more care actually being delivered. So what presumably it means is like, I think humans are awful at sometimes missing things, right? Like the human brain is only so capable of like actually potentially capturing what's happening in real time. And so it leads me to believe that, you know, I think without the care being changed or more care being delivered, that there was just undercoding to your point on the half of providers.

22:55Michael Esquivel: But so in that way, and we know these provider systems are struggling, right, from a net profit margin. Like these are not healthy businesses. So I'm okay with honestly, them getting the proper reimbursement. It was actual upcoding. That's an issue. You know, I think this is to be expected, right? I think this is like era one and it is highly inflationary and there's going to be some bad headlines. What I hope AI ultimately gets us to, honestly, is get rid of all this crap and allow for real-time payments, right? Where like the AI, because of the rails that are being built, because data can flow and logic can happen so quickly at a superhuman level and, right, in real time that you actually can check the eligibility, the provide the service, build for the service, you know, run the prior auth all without having all this back and forth that we used to have and just strip out the friction in the system.

23:49Michael Esquivel: So ultimately, like rather than have a put another way, a provider bot and a payer bot, it's kind of like one rail of which a real time adjudication settlement happens. That's my hope. Right. And I think there's a lot that has to happen, but that could not happen beforehand because software was so static. Right. Now that I think this software is truly dynamic real time and can automate a lot of the frictional way. I think there's a hope there. And then I think that'll decrease a lot of the spend and the administrative spend that we have in the RCM part of the healthcare stack, which is a lot of paper and back and forth and humans and the like.

24:26Michael Esquivel: Like, let's just get rid of that. I think AI provides that opportunity. Maybe not here today, but I think here pretty soon. And that'll drive down this cost is my belief.

24:36Steve Kraus: Yeah, Dr. Oz certainly highlighted a similar theme. The other thing he said, which was interesting, and I think this is right, is ultimately the answer is to change the incentive structure, right? He specifically pointed to ACOs and the broader shift away from fee-for-service. And I think your real-time adjudication plus setting the incentive structure in a way where we're getting quality outcomes and fee-for-outcome type scenarios is going to really, I think, be the drivers collectively of turning this from an inflationary event to a truly deflationary event while giving us all better care along the journey as well.

25:16Steve Kraus: So exciting.

25:17Michael Esquivel: Yeah, totally, Michael. Totally, Michael. I mean, the administration, you know, they could have chosen to, and they might develop AI fee-for-service codes, but their first initiative, big initiative was access, as you know, right? The access initiative. And that, you know, Jacob Schiff, who runs that, I've spent a lot of time with him, like his whole point is we did not want to reimburse activity, right? Which would be a normal fee-for-service code. We wanted, to your point, reimburse outcome. And so the whole access model, it looks like a value-based kind of care model. There is an upfront payment to reward the software providers or AI providers for the effort they have to put into developing.

25:52Michael Esquivel: But then a lot of 50 % of the payment is based on outcomes. So yes, I think that is the way to remediate this as well.

26:00Steve Kraus: Yeah, it's going to be interesting. It's an exciting time, Steve. We're so privileged to live in this moment that we're seeing in healthcare.

26:10Michael Esquivel: It's for sure wild. And there's lots of opportunity, but lots of challenges too. Okay, speaking of uncharted territories that we're operating in, the final story today, Michael, is a kind of, I think, a paper that might've been overlooked by our industry. It's a wonky paper. It's a white paper that the Digital Health Center of Excellence, the FDA's center that's regulating this AI, both for biotech as well as for provider payer world. They put out a paper called, it's a thrilling title here, Michael. It's called Considerations for the Regulation of Generative AI-enabled Medical Devices. But I think it's an important paper that we talk to our constituents about because actually comments are open until October 19th.

26:53Michael Esquivel: And I actually know Jared Seehofer, who runs this agency, is the new head of CTO for AI at the FDA. Big role that he has. And he very much wants our industry to comment on this paper. So please, listeners, if you're building, if you're operating in this new AI world in healthcare, please take the time to read the paper and comment on it. But here's a couple of interesting things I think that they're thinking about. By the way, there was no law, no regulation promulgated. This was just a paper that asked a bunch of open-ended questions. But some things that I think are interesting is, one, they're thinking about how they regulate these devices, meaning let's just keep it simple, kind of AI doctor in your pocket.

27:30Michael Esquivel: There's lots of examples, but let's just use that. And so one way they're thinking about is that they're actually thinking about how they treat these models like clinicians, that you get medical training and then you become a resident and then you become a fellow, right? And then you become a chief fellow and then you actually become an attending. And along the way, you get more and more and more of a leash to practice on your own independently, right? And then you also have to renew your medical license and take the medical exams and get the proper amount of CME. So they're actually thinking about it in that context, which I think is smart and really interesting.

28:05Michael Esquivel: Unclear how they're going to do that, but that's one. The second axis is they realize they have to think about how severe is the case of the disease or the patient that the AI is treating. Like how critical is it? Like if it's in the ICU, hey, maybe we have different regulations for that than if it is primary care, for instance, where it might not be as risky or renewing prescriptions that are low acuity prescriptions. I think that's interesting. And then the final one, which I'm very curious about you from a liability regulatory perspective, is they have to think about, you know, a lot of these vendors are going to be building on top the foundation models, right?

28:41Michael Esquivel: So you have one piece of software, which is actually the interface and the workflow and the actual product itself, right? AI doctor in the box, let's just say again, the council doctronics, you know, of the world. But then those vendors and lots of others are building on top of OpenAI, Gemini, Claude, you know, maybe some open source models, right? And so what happens when the underlying foundational model gets updated? Who's responsible for that? Is it the foundational model company or is it actually the software vendor, the Doctronics councils, you name your front facing solution that is building on top of that model that owns that risk and liability?

29:18Michael Esquivel: So they've been asking that question as well and they don't really have any answers, but a really good framework for people to read. It's really pertinent in terms of the timing of this podcast. Please read it. Please comment if you have comments because it's going to be critical. This is like, by the way, this administration wants to move within two years to create this regulation. I know that. So this is not like this 10-year regulation. This is happening here and now. But Michael, I'm curious about the last axis that they talked about, this foundational model liability versus the actual vendor themselves.

29:46Michael Esquivel: How do you think this shakes out from, again, the legal liability perspective of the world?

29:52Steve Kraus: Yeah, great question, Steve. So maybe taking up leveling this just a little bit, as we all know, to submit a product to the FDA for approval. The product has to be both safe and effective, right? And so the question for these technologies is part of the product is the underlying model. I just want to orient us that way. So you've got these, the real tension besides who owns the update is these foundational model companies, they don't want to share confidential information with these downstream companies in order for their overall product to get approved by the FDA, right? And so it raised, I saw in that, it raised this concept of a master file, right?

30:32Steve Kraus: Yeah. Which these foundational model companies would enter into with the FDA, where it would share those technical details with the FDA directly, and how the model was built, how it was trained, how it was tested, et cetera, without having to hand that information over to the downstream companies that are building on top of it. And so I think that, you know, it's sort of roundabout to answer your question is, you know, is that the approach that's going to be taken? Because otherwise then I don't know how some of these technologies, at least in the framework that is being laid out here in this piece, I don't know how you get a product if you're building on top of a model to get approved through the FDA without understanding the underlying model that is the basis, a big core basis of the product itself.

31:19Steve Kraus: So I think that's going to be a very key, important threshold question. And then your follow-on is the really open one, which is, so updates, changes, I mean, these models are evolving by the millisecond. Who's responsible? Is this this master file concept where they have to constantly be updating the technical details with the FDA? Is it the responsibility of the downstream company building on top of it? You know, it's a big question. So I think, Steve, your call to action for our ecosystem to get Jared all of these comments and feedback, I think is going to be important because I don't know where you draw that line and who's going to bear that risk.

31:58Steve Kraus: It's a great question. I think it's going to be the big legal question when it comes to getting these health tech, digital health technologies approved by the FDA.

32:08Michael Esquivel: Well, you know, Michael, you said it's such a dynamic, fun time. I mean, you think about the FDA, even Jared, right? Like back in the day, you talked about FDA's responsibility, safe and effective. And you used to have an old device and, you know, you'd submit, you'd get a 510K, and then maybe you'd update it two years later and you'd go back and get like a, get addendum to the 510K. Now, to your point, these models are changing on the millisecond. Like how does the FDA, like, you know, how do they think about that? I mean, it's the whole world is just including the regulatory frameworks are just changing rapidly.

32:36Michael Esquivel: So people, please do comment. What a great podcast. I mean, we had some technical errors, but we talked a lot. Hopefully our listeners won't know, but we talked a lot about some great technical content and something matter. And by the way, Michael, I love Halle. She's the best, but I needed you as the co-host on this podcast because we got pretty legal pretty quickly here, my friend. So thanks for covering such a great set of stories today. I hope our listeners enjoyed.

33:00Steve Kraus: Yeah. Thanks. Thanks, Steve. And congratulations again to you and your partners at Bessemer on the amazing new fundraise.

33:07Michael Esquivel: Appreciate it, my friend. All right. Happy October, everyone.

33:37Michael Esquivel: Visit our website, heartofhealthcarepodcast.com, for show notes and details.