Signature Series: No Corporate Slides. Real Strategies and Resources to Deliver Purpose-Driven AI | LSI USA '25

Armen Vidian  0:05  
All right, you all set. So first of all, I want to thank Henry Peck and the LSI team for inviting me today to sit with the two most exciting CEOs in surgical AI today, truly, we are at an exciting moment for AI and healthcare, unlike even just a year or two ago, when AI was viewed by many hospital systems as an interesting add on to medical devices. Today, we truly see such as from both of these companies truly AI, native products that have impact for patient care, the physician experience and the medical device industry writ large. So this is a very exciting panel, and time to be here today. So before introducing myself and my firm. I will hand it over to Ann first to introduce Moon surgical, and then same for gate.

Anne Osdoit  1:06  
Yes. Thanks. Hi everyone. My name is Ann. I'm the CEO at Moon surgical. I'm also a partner at Sophie Nova, partners specifically in charge of the MD start Medtech venture builder, which where Moon originated from.

Gabriel Jones  1:22  
Thank you. Honored to be here with this esteemed panel. I'm Gabe Jones, the co founder of propria, which is a Seattle based surgical data company, the first real time AI data company in surgery, we would claim, but we can battle that out as well. A lot of respect for what you're doing at Moon is a surgeon invented and driven platform that's collecting a tremendous amount of of data about the surgery itself. And we will get into our thoughts and strategies about how to use those data for many applications, first and foremost, to benefit the surgeons in treating the patients.

Armen Vidian  1:57  
Excellent. And I'm Armen Vidian, the managing partner and co founder of Recode ventures. Recode ventures and invests at the seed and series A for companies at the intersection of artificial intelligence and healthcare, and we believe we are the first firm solely dedicated to that intersection. At the early stage, we had seven companies in our first in our first fund, and we are getting ready to raise our second fund later this year. In full disclosure, we are all in also investors in Proprio. So I'll get started with, first of all, for both of you, when you talk to your customers and to physicians and so on, what do they believe that AI is going to add to their experience, to patient care, and are there certain things they are anticipating it will yet add that they haven't yet seen?

Anne Osdoit  2:58  
So that's a great question. I think when we speak to surgeons, you know, AI is, is a brick, right? It's a technology. It has to be congruent with your value proposition at a broader scale, and it has to serve basically the purpose or the problem that you're trying to address as a company. So we are developing a surgical assistance platform in the field of soft tissue surgery, and what we are specifically trying to solve for is access, right? Really, developing a platform that can be used in, Nur for any type of laparoscopy, and that can deliver some of these robotic features in a way that integrates a lot more in the current way of doing surgery and laparoscopy, leveraging the existing tools, workflows, processes, et cetera. So when we speak to our customers about AI, it is about making it more accessible, which means in our target market segment, which is high volume, low acuity settings, right ASC, hospital outpatient departments, it means making it easier and more efficient. So for instance, using AI to automate the setup of the system and make sure that it's perfectly set up. Not only you know, for that procedure were, for that clinical indication, for that surgeon in that room, based on all the procedures they've done before and what we can learn from it, it means, you know, helping them manage some of their instruments in the way that, you know, mimics their preferences, if they were to do it by hand, so that they don't even have to worry about that, and it saves them time. So it has to be within your broader value proposition. They get excited, of course, because you have something that is both a lot broader in terms of access, but also a lot more tailored and personalized in terms of its behavior. So kind of winning on both. Sides,

Gabriel Jones  5:00  
yep. Can I echo that and just sort of jump? I think it was like a near mid and long term role of the CEO of a company like moon or Proprio, where we are translators to clinicians, and then a series of different users and user customers over time that will evolve. I think that's maybe something we'll talk about little bit later, but that's how I look at it is right now, being surgeon driven in the platform that we've built is all about making it easier for them to make the right clinical decision, faster, better, faster, safer. It really is about that. And I echo what Ann said, this is like the difference that we would say, between navigation and guidance. And so traditionally, navigation for surgery is a couple cameras, tracking markers that are either drilled into the body or place on the body in some way, and even robotic systems, many of them that are image guided, will will look at those markers to find the patient as soon as you tip into AI, actually evaluating data sets and feeds and providing guidance into a different clinical decision, maybe based on real time measurements of that surgery and the KPIs of success for that surgery in that patient, that pathology, you tip into just a different category of product, which is guidance and the trust factor with a surgeon needs to be, obviously, very high, even with navigation. But now, if we're saying, do this instead of what I see you doing right here. That's just another even higher bar. If it was 99% now it's asymptoting to 100 right. And so there's a translation and an explanation and an education that we get to do standing side by side with the surgeons. And there are these magical moments where they can actually see yellow went to green, right? That simple what's happening under the hood. Some of them would like to know that. Others just want to be able to trust it that helped me to take care of that patient in that day. So that's how I view the near term. I think we'll get into the mid and the long where there's other opportunities in efficiency and helping other products and companies to perform at their very best, verifiably with data. And I think there's some platform considerations about whether to be an open platform or closed. And in the past, I think in med device, it would have been very much about acquiring implants and pulling them onto the platform, whereas some of these new technology opportunities, especially with AI, provide an opportunity to open the platform and benefit many, many more players, and thus, customers and partners. But I think that's foreshadowing some of the other things we're going to talk about. You know,

Armen Vidian  7:23  
it's interesting to hear in both of your answers, you talk about things like efficiency, clinical decision making and so forth. When you talk to your customers at the end of the day, in the hierarchy of needs, what are they buying first? Is it clinical superiority? Is it safety? Is it speed? What, what first comes up with them and they're in your conversations,

Anne Osdoit  7:49  
it all depends on their pain points, right? Which themselves depend on their market segment in a way, right? So, for instance, we're going for the low acuity procedures, high volume settings, which are places are not teaching hospitals, and so they are driven by efficiency. Of course, they have massive staff shortages, and they have specific problems around allocating the right procedures to the right platform, allocating the right staff and resources to the right platform, etc. So, you know, they're extremely sensitive to quality of life of their staff at work in order to retain their staff, they're sensitive about agility and how they can deploy their staff as you know, they're there, there's a high reliance on temporary workforce, and then there's sensitive around, you know, the economics, of course, of the platform and an overall efficiency right? They are tracking all these metrics, you know, wheel to wheel time, incisions, you know, and skin to skin time, turnover time, etc. So you need to show them that you are delivering benefits. And you know, AI is, of course, one of the vectors to do that, but you're going to be cost neutral in the end. And so it is about really building that conviction process and then complementing it with your return on investment model, and that is ultimately what drives the sale. It's a bit of a convoluted process. Ai, is really only one piece of it. It is the enabler of that greater efficiency.

Gabriel Jones  9:38  
Again, I like what what Ann said, there's room for, we're at the very beginning of this cresting of this first wave, so there's, there's room for a lot of different strategies that could succeed here. In our case, we're very focused on the hardest procedures that are currently unsolved. So it's a big swing in that respect. I would say Moon is also but targeting a different part of the of the market, which is a huge opportunity too. So for us, we target. The top academic centers in the world. So think University of Washington, Duke, Ohio, state, Columbia, these kinds of places where they are teaching institutions, and they have rate limiting factors for how many surgeons can be trained in a fellowship program per year, and the type of information that can be provided to those fellows as they go through a program, side by side with maybe four, six or 10 world class surgeons who are doing really complex spinal surgeries like scoliosis corrections and things like that that, you know, there isn't really good real time data for so in the chart of like, technically difficult to solve, clinically very difficult to solve, we've picked the very upper right corner, which is has elements that are emblematic of everything else in the chart, but they're the hardest places to go, and in classic disruption, that's kind of where we start with a beach head like that. So connecting a data platform that surgeons are really looking for the answer to your word is your question is one word which is certainty. They want, certainty of success that their technique is leading to the best possible outcome, and that when they go and teach that to a fellow who's standing right across from the them in the operating room, and they're giving them feedback, it's not just go faster, do better, that's not really helpful. It's actually quantifiable. It's juxtaposable. You can compare it against another data set, and say, with certainty, this many pixels more should be scraped away from the surface of that bone. I mean, that's the level of power of these technologies now, and we're bringing them in the operating room to provide that certainty. So our job is to translate what the technologies can do to provide that certainty.

Armen Vidian  11:33  
You know, one thing I've noticed in talking to physicians about AI and in medicine, as opposed to seven, eight years ago, when I was first investing in companies like caption health or AI guided cardiac ultrasound, was questions back then were, will it work? Will it very fundamental right of the getting over the credibility hump today, it seems to be much more subtle, less about if, but rather what specific applications, how it fits in with my workflow. A lot more subtleties. Are you seeing the same or are there recurring themes?

Anne Osdoit  12:08  
I think I would agree with that. It's a lot about productizing and then ultimately monetizing those products. It is, of course, about getting them to regulatory agencies, which, you know, we'll speak about, not, not all products have to go through regulatory agencies, but in terms of the technical capabilities, we're absolutely there, right? I mean, there's no question about that. You know, we can predict procedure end times and enable dynamic scheduling. We can do inventory management. We can help with, you know, instrument movement automation. We can, of course, automate the setup of our robot. I mean, there the what we can do is actually pretty sophisticated and infinite. What we want to do in terms of delivering products and specifically getting some of them through the agencies has to be a lot more deliberate, as it has to feed your value proposition.

Gabriel Jones  13:04  
Yeah, again, I would echo what Ann said, but I would add to that, it's getting so good at predictive that we'll be able to say not just when is the case going to be or the surgery going to be complete within a matter of minutes. Instead of a charge nurse watching a video saying, Oh, the patient's off the room, let's start to turn it over, which is spoiler alert, how most hospitals operate most of the time. It's such a valuable space. The operating room, you know, in many North American hospitals, is $200 a minute, and it's 60% of their contribution margin for the hospital, and $2 trillion a US healthcare spend flows through the hospital. So it's very it's a high ROI. If we can figure it out, we can do even more than predict the outcome of the case. I'm sure you guys have a bunch of insights there too. We can also say, Can we redefine outcomes? Can you be predictive within the surgery so you can make different guidance driven decisions that will lead to better clinical and economic outcomes and verifiable with a number, while you still have an opportunity to change that outcome? That's that's the reality we're living in now. And then, going back to your earlier question, thinking mid and long term, I think it's incumbent upon us as ambitious companies, to take our platforms to other partners who can help us bring down the acquisition costs. Robots are expensive. Ai, development is expensive. We have to go get talent from, you know, various other companies, not just in med tech, and those people are expensive, that expertise is expensive. So we have to invent ways to build valuable relationship with other companies, be they implant companies or other software companies that benefit from the data that we're able to collect and and we have to figure out those relationships so that we can drive down the acquisition cost and then bring these technologies to bear writ large.

Armen Vidian  14:43  
And that brings me, actually, to my next question, and a question that is near and dear to my heart for some time is, how can AI fundamentally change, if at all, and if so, how change the medical device industry? What does it mean? To us as innovators, as investors, as operators and medical device companies, big and small. What? What does aI mean to how we run as companies?

Gabriel Jones  15:13  
Okay, I have lots of opinions about this. We only have 14 minutes, so I'll try to do it really short. I think one of the most controversial things that is stated at conferences like this, is, what is the role of the rep, and can we get the rep out of the room actually take the contrarian approach to that, which is, what's the job description of the future rep? How could they do more and different things, as opposed to bringing, you know, 200 spinal fusion screws into every surgery, because you know that they had to keep in their trunk, because you don't know what that surgeon is going to ask for on that day. That seems like a misallocation of resources and expertise. So I view that person and their five years out job description as being really exciting, more data oriented. That person may have two or three different customers within the hospital. Some are surgeons, but some are beneficiaries of the data and analytics and the platforms that we're building, and they should be expert in that, just like today. If you if you asked who's winning, it's fine, most people would say Globus, because they did one really smart thing, which was integrate and then they taught their metal reps to sell robots and navigation and imaging and operate everything. That's great from a business ROI perspective, but I think we can do even better, which is train the next generation and enable them, treat them a rep, like an internal customer, if you will, and give them tools and weapons to go after these unsolved problems. Right now, that's really exciting. So I actually take the opposite view, which is, there'll still be people in the operating room doing various different things, but they'll be much more efficient with tools like moon and Proprio, and their expertise will actually extrapolate from the or throughout the entire building of the hospital, impact, revenue, cycle, management, these kind of areas. And I can tell you that that transition has already started, which is pretty damn exciting,

Anne Osdoit  16:58  
yeah. And I think as an industry and, you know, going back to what you were saying earlier, it is also going to change the way we view our development plans, the way we view business development as part of our activities, you know, if you're developing an intervention, you know, okay, for us, for instance, we build hardware, right? We spend two or three years of the company building hardware, we hired people who knew how to build a robot, right? And we built a first generation and a second generation, and both got tiered by the FDA, and now we have a robot. It's great. We can reduce the cost, we can reduce the footprint. I mean, there's all sorts of things we can do with the hardware, but the future of the company is really now about leveraging the data that the robot is sensing in the or so, then you become thinking, we you start thinking, Okay, do I have the right resources for that? They're not the same people in the company. Are we becoming a data company, and as what we can do with this data is actually pretty infinite. Should we be building everything ourselves? Of course, the answer is no. And there are many partners out there, companies, you know, labs, different initiatives within healthcare and outside of healthcare. You know, Nvidia being an example, of course, where there's a lot of things we can leverage and integrate and people we can partner with. So you start thinking, Okay, this is not now just a task for the R and D team. It is actually a broader business development effort to try to understand what's core and what you're going to be aggregating and integrating from other people. And I think you know, for anyone who is developing, you know, an intervention or a technology that is used in some sort of procedure, it makes sense, right? You are going to develop a ton of data based on how your device is getting used, and ultimately, it's up to you what you make of that data. But there are a lot of possibilities, and that's opening an entire new value chain, if you will. Yeah,

Armen Vidian  18:58  
are we nuts to surmise that the device industry has an opportunity, with AI to become a lighter weight industry. And by that, I mean, you know, to investors, one of the scary parts of investing in devices is that people come to say, Okay, we're gonna have to do this clinical trial, a PMA study, right, many years and millions of dollars, and then, by the way, we're gonna have to go to market. We're gonna have to hire two reps per account, and people's, you know, eyes start to glaze right when you look at these kinds of numbers. Are we going too far to say that AI can help us reinvent that model in ways we haven't thought of yet?

Gabriel Jones  19:37  
Absolutely not. Sorry, I jumped in there. But if we talked about this a little bit last night, but I think the the metal company, the implant company, needs to fundamentally change where they make their investments, and we need to enable them to do that as well. And that means we need to have a diverse set of hardware, software, systems engineering. Even within AI there's all these different categories, from two. To science and machine learning, to deep learning and the newest models and rag and raft, and it goes on and on and on, and we need to have those folks working on major projects internally, keep them motivated by the mission, but also the technical goals we're going after, meanwhile, helping to actually impact the bottom line for the implant companies out there too. And that means that their workforce needs to change quite a bit. I mean, you're seeing this sort of ripple through MSK surgery and spine in particular right now, with Stryker selling off their implants, really started with invasive being acquired by Globus. And so there's a real need for an emergence of several companies that approach the New Economics of healthcare differently, and that means the resources and the margin are going to shift around, and the companies that are better prepared for that have been thinking about it for a while. Well, obviously we're biased, but we think a certain set of companies are much better prepared for that than others,

Anne Osdoit  20:53  
and I think AI can actually help us be a lot more efficient in a way, right? We, you know, there is this interesting parallel about, you know, basically the value of synthetic data and what it did to the self driving car industry, essentially getting them to a level of safety that was good enough for these cars to become products, and, you know, be turned into services. I think this is also applicable to our industry, right? And so we need, I think, to be kind of spearheading the efforts to educate the FDA, to educate, you know, people around us about the value that generative AI can be for developing these synthetic data sets. And then we can actually, you know, train things in a way that is a lot lower bar in terms of, you know, finding actual data from procedures and ORs and etc. It is a discussion that we are currently having with the FDA. I mean, I'm not telling you they're fully educated on that, but it is certainly, you know, making its way and, you know, when we get there, I think we will also be a lot more efficient in the way we develop these new features. Yeah, I

Armen Vidian  22:07  
couldn't agree more. Ann, it seems that from my time in the industry, when I was in transcatheter valves, there was opportunities from everything from, you know, doing a case, incident reporting, to your documentation, from sales, to all of the tracking that we used to do, every one of these seems to be an opportunity to operate in a more lightweight, more efficient pattern, which reminds me next question is, how do we need to think about selling AI native products a little bit differently from conventional medical devices or traditional devices? Do you need a different sales force. Do you craft your messages differently into different people in the ecosystem? Are there other things to think about and just how you position and sell your products that are that might be different from what people are used to?

Anne Osdoit  22:56  
I think you're selling value right to clinicians. And you know for you it's like better decision making, better precision, or for us, it might be better efficiency, et cetera. I think you so that part doesn't really change, but what you need to navigate are sensitivities around connectivity, around data ownership, around cybersecurity. I mean, all of that for sure, requires a certain degree of training for your sales first, because these questions will be asked, these things need to be in commercial agreements and and for now, it's not entirely standard. So, you know, a lot of people are having those discussions for the first time, and need to be, you know, handheld as they're as they're having them. So, yeah, for sure, some additional skills. But you know, the fundamental value and search and conviction that you're building is, you know, something that any rep should be able to do. A

Gabriel Jones  23:56  
year ago, we talked about the AI moment we were in May of last year in Chicago. And the answer was, there's a lot of the aperture is open. Everyone's listening. If you have a cool prototype or proof of concept that'll get you into the conversation, well, congratulations. It's a year later, almost, and that's not enough now, because anybody in their garage can build a pretty cool prototype by using foundation model and maybe a little bit of tweaking for the inference if they have access to any kind of a data set. So now it's about solutions. I agree with what Ann said, and the cool AI moment we're in right now is there a lot more solutions and models out there that can be the right tool for the right job. So you can take a surgical video and multi modal data set and saying we can generate a nearly perfect automated operative note that is something that can be exported into the EMR and directly into to RCM, and shorten the payback cycle well that can drive the flywheel of pre authorization, or automated prior authorization. These things are actually somewhat possible, as opposed to just conceivable. Or prototypable now, and I agree with what Ann said, it's all about value, so we've got to pick the right things to build that are very conspicuous in how valuable they are. But I'll go even further and say the ability to simulate other products for other companies with both real and simulated data is a massive ROI potential impact, because that development of productization and commercialization is very different, a different path. We need to work with our partners at FDA on that. But there's even a third category of development and commercialization that can run even faster. That RCM product I just mentioned is not a 510, k, right? It starts with data from the operating room, but those engineers are running at a very different pace, with a different set of rules. And I think that's important to call out, because the modern medical technology company, or some of our friends, would say tech med is is both or all three of those things at once, which is hard to do, but if you can execute on it, there are multiple monetization strategies, both in the or in the hospital and even outside of

Armen Vidian  26:05  
it. And what have been your experiences in working with the FDA on AI native products? Do they, as far as the regulatory path goes? Are they open more to 510, K or de novo 510 k pathways, as opposed to PMA route? How are they in thinking about things like new releases that you might come out with and so forth. What's been your experiences?

Anne Osdoit  26:28  
Yeah, so we, we just announced, two hours ago, a new FDA clearance for our first AI based intraoperative feature. So really controlling the movement of the scope using AI. It's, you know, it was a two year process with the FDA, but it is, it is a 510, K, I think, I think it's a lot about educating the FDA and reassuring them, and also taking baby steps, right? And so really, you know, yeah, slicing it in a way that is acceptable for them to swallow. The FDA likes design frozen, deterministic software, you know, they're a little uncomfortable when things might evolve and performance might might evolve over time. Arguably, they should be right. And so they have this new modality that is called a predetermined Change Control Plan, which is meant to define the triggers that will allow you, essentially, to modify your algorithm based on, you know, a drop in performance based on having a large chunk of data that you're adding to your data set based on wanting to, you know, do something slightly differently, basically, from your for your original algorithm, I would say that the V, F, D, A itself is a little bit in calibration mode these things, right? So providing valuable feedback, but also sort of uncertain as to how much they can accept, but with a lot of dialog, you know, I think, I think we will get more and more products cured. Yeah, for sure.

Gabriel Jones  28:16  
Congratulations. By the way, that's awesome news, and it is a rising tide, right? Right. So every success that a company like Moon surgical has in working with the agency, you know, that benefits Proprio, and I hope vice versa. We are currently under a media embargo, so I can't say anything else regarding any announcements that may happen very soon for Proprio, but I think there's a lot of good stuff that's that's happening, especially with FDA. It's been a turbulent time, but working closely with them on a sequence of algorithmic and AI driven improvements is the is the key to this game, and continuing to educate and respond. And it's it's challenging, and sometimes we have to go a little bit slower just to be very intentional and methodical there. But separately, we can go really, really fast outside the operating room, and that's where we need to continue to have those big challenges, where we can keep attracting talent from even the apples and the metas and up where we live in Seattle, the Microsofts of the world and the Amazons to come tackle these big swings that are worth it for them to lean in on

Armen Vidian  29:20  
excellent well, we just have a few seconds So real quickly between the three of us, how do we see the medical device industry different? Because of AI, 10 years from now?

Anne Osdoit  29:33  
Yeah, I think you know, any medical device, as I said, is, you know, just by the simple fact that it exists and is being used by patients is, you know, a Data Hub, right? And will benefit from being connected or communicated or having some sort of, you know, analysis performed on that device or intervention. So I think it's going to be massive and everywhere. It's going to take time, obviously, to get there, but I. Yeah, you know, the value of that data is in making it more predictable, more personalized, more accessible, etc. And you know that's that's not going away.

Gabriel Jones  30:10  
Okay, let's close with a provocative statement. Within 10 years, you said one of the largest players will be dead, one of the largest players that tries to bone graft on AI to their business and can't figure it out, will get disrupted by one of these companies like ours, I hope, and that'll change the entire industry, because everyone will have to react to that. They'll have to change their sales force, they'll have to change how they hire. They'll have to change how they communicate with the street, really fundamentally altering the way that we think about the potential outcome, and even define it for surgery, both clinically and economically. I think within 10 years, we'll see that. I don't know which one it'll be. Ask me again next year at LSI, but, you know, I have a hunch.

Armen Vidian  30:52  
So obviously, I'm a biased source, but I 100% agree with you that a major player, if not acting quickly, is going to find itself playing a big game of catch up that may not work. We completely agree. And from our standpoint, we see AI reinventing the surgical experience into one that is elegant in the way we expect to see from our iPhones or consumer devices that it should go into the surgery knowing exactly what's going to happen in a seamless experience with technology that the physician actually looks forward to using, the patient has much better outcomes as a result of it, and the industry can run in a much more smooth and efficient way, in ways that we haven't even thought of yet. But somebody in this room and certainly present to us in the years to come anyway. Thank you very much. 

Anne Osdoit  31:45  
Thank you. 

Armen Vidian  31:46  
Been a pleasure.