Full Transcript Introduction: Dr. Justin Smith 0:01 Welcome to the Startup Physicians podcast where we explore how doctors 0:06 lead healthcare innovation. Please like and follow our show to join our 0:10 community of physicians contributing to meaningful change. 0:17 Welcome back to the Startup Physicians podcast. This is your host, Dr. Allison 0:20 Kerman, and I am joined today by Dr. Justin Smith, an orthopedic surgeon and 0:26 AI guru who's been building his own platform and has so many interesting 0:31 things to share about the journey and and the build. So Justin, thank you so 0:35 much for joining us today. >> Yeah, thanks for having me. 0:38 >> I really enjoyed hearing about your background in our conversation. I'd love 0:42 if you could share with people the arc of your career and and how you went from 0:47 being focused on orthopedics to what you're doing today. 0:51 >> Yeah, absolutely. grew up in Atlanta, University of Georgia, genetics major. 0:55 So, started out as a nerd and then trained in Michigan, orthopedic surgery, 1:00 and then orthopedic sports fellowship in Steman Hawkins in Greenville, South 1:04 Carolina. And it was there that met my two co-fellows, Nate Bose and Todd 1:08 Walker, where I really kind of started to get that entrepreneurial bug. And 1:12 Todd was really kind of the impetus of it. He out in California during his 1:16 residency had already started developing some patents and products and he was 1:20 very energetic guy and helped us to start the idea of building a coffee 1:24 company called morning rounds coffee. It's similar to black rifle except we 1:28 donate 25% of our proceeds back to healthcare workers. And so we started 1:32 that from scratch all bootstrapped and really learned a lot about you know 1:35 creating a business all together. It was a passion project we've won awards with 1:40 it. We're now in University of Cincinnati and and that was a fun 1:43 undertaking that was just learning about the intricacies of building a business. 1:47 Since then, I've developed medical products that I've gotten FDA approval, 1:51 other small companies and then went into practice for a massive hospital system 1:57 fresh out of fellowship. Clinically, loved it, but you know, as many other 2:00 physicians experience, very little autonomy. And so, you know, December of 2:04 last year, decided to leave and start my own private orthopedic practice. and in 2:08 doing so realized that there's a lot of opportunity especially with the AI 2:13 revolution that's coming to build an infrastructure that would help leverage 2:17 that technology to reduce the overhead to run a practice and I knew that I 2:21 wanted it and others would want it. So I ended up creating a company and we're 2:26 we're pushing as hard as we can now. >> Congratulations. I think there's a lot
Why AI is empowering physician founders 2:29 of physicians who are feeling really empowered by the AI revolution right now 2:35 that software engineering is now accessible to those of us who are 2:40 traditionally more considered subject matter experts and not coders. And I 2:44 think that sometimes being able to be a visionary and to identify well these are 2:49 all the problems and this is potentially how I could create a product to solve it 2:54 and then actually having access to being able to do that. it really can get 2:57 something going. >> Absolutely. Nowadays, the technology 3:00 offers you this huge opportunity. Knowledge no longer has capital anymore. 3:05 Knowledge is accessible to everybody and that's the beauty. The keys of the 3:08 castle now are amongst the masses. So information about business, about 3:12 medicine, about technology is more accessible than ever. It it behooves us 3:16 to be able to leverage that technology to be able to do all these beautiful 3:20 things. I do lecture internationally on trying to help you know individuals 3:24 understand how to use this technology best practices to help them build all 3:29 these great ideas that they have and then how to go about it in in the best 3:33 way. >> That being said there is a difference 3:36 between something you can vibe code in an afternoon as a non-technical person 3:40 and what it takes to actually build production ready healthcare software. So 3:45 it definitely does feel very accessible these days. But then there's also the 3:51 point of if you don't have experience overseeing product development, you can 3:55 go down a lot of rabbit holes. You can be unaware of different compliance 4:00 requirements and it still requires technical expertise to get 4:05 productionready healthcare software. I'd love to hear what your journey has been 4:10 with a development team and and finding the right resources and understanding 4:14 those the scope of building something like what you're building. 4:17 >> Absolutely. The most important thing is just having humility in this process, 4:21 right? You don't know what you don't know. Uh AI is good and it can tell you 4:25 your blind spots and you should absolutely leverage that to help you 4:29 understand what you don't know, what you should be thinking about and when. It 4:33 was actually during my three-year-old's birthday party that I was telling him 4:37 about this idea and how we're going to automate clinical practice. And my 4:41 neighbor is a software engineer. He's like, "I'm a machine learning scientist. 4:44 Like, do you have anybody running your software?" I was like, "No." He's like, 4:48 "I'll do it for you." And it was actually through him because I I knew 4:52 nothing about software engineering, the process and so forth. I think AI is a 4:55 great tool, but you don't know when it's erroring unless you're an expert of 4:59 that, right? And so, I think there always needs to be a human in the loop 5:02 to vet the information that you're given. Otherwise, you just take it and 5:05 hope that it's right. With Chris, he's been able to really teach me the process 5:09 of developing it, right? So you know from feature road maps to actually 5:14 building the technology to QA testing to then putting it into testing staging 5:19 production how does that whole flow work what are the different you know types of 5:24 code and so forth how does the database run how is the user interface and then 5:29 you know unfortunately doing a healthcare app that touches PHI it's not 5:32 like a fitness app right if this was a fitness app we would have been in the 5:36 market in a month but with healthcare because of that security we had to a
Healthcare security, PHI, and SOC 2 compliance 5:41 high level of compliance and probably the most authoritative compliance is 5:44 SOCK 2 compliance which is not only that your data is secure but how does your 5:48 company run in order to maintain that that level of security. 5:52 >> Yeah. And it it's a lot. I've seen software development projects go 5:57 sideways when people who don't have an overview of the feature requirements or 6:02 the product roadmap or what's going to be developed when just have a lot of 6:06 ideas of what needs to be included in the very earliest versions. And every 6:10 single feature needs to earn its place on the product roadmap strategically. It 6:15 has to be blocking something or it's unlocking something that is some sort of 6:19 value. But if you spend all of your time adding a ton of features that are nice 6:26 to haves and you don't actually get something into the market, you have the 6:30 risk of like never actually releasing it or releasing something that that doesn't 6:34 work. >> Yeah, you're absolutely right. You know, 6:36 Chris was actually very helpful to mentor me in that piece. He had built 6:40 different software solutions for different industries and he would always 6:43 say yes I'm the subject matter expert and that helped but at the same token 6:47 like what is whimsical and I would really want this for me may not pertain 6:51 to others the ROI is not really there and in in doing so I understand that my 6:57 viewpoint is very narrow and so what's been cool is not only from Todd and 7:02 Nate's perspective but our investors are all orthopedic surgeons and different 7:06 clinicians and they are all beta testing our product and so a lot of what is 7:11 being built is being directed by other physicians, right? If I want it, I can 7:15 assume other people want it, but I want to hear other people tell me and 7:18 prioritize what gets built when and then how does that function?
Why physicians still have an edge as subject matter experts 7:22 >> Yes. And I I want to call out something you said about being like a subject 7:27 matter expert. A couple months ago, I was at this retreat. It was like a 7:31 hackathon with like a ton of developers and then like three doctors that came 7:35 with me. It was interesting to have conversations with these developers 7:38 because they felt like they were like our world is now accessible to anyone. 7:43 They were talking to us as the physicians, but you know where the 7:47 problems are and you know not just a description of the problem but like all 7:51 the edge cases of it like every single aspect of it that all the features that 7:56 are required being the subject matter expert on that thing. it it's still 8:01 great to come together with developers, but their world is actually accessible 8:04 now. And so, one of the things that I'm always trying to do is to find the right 8:09 combination of skills and talents that are needed on different initiatives or 8:12 ventures. And I certainly think that really great developers who can work 8:17 handinhand with the subject matter expert and really guide the process and 8:21 make sure that things are being done with high fidelity and and compliance is 8:25 so important because I actually don't think that we as physicians I don't 8:29 think that our biggest contribution is going to be in actually coding like I I 8:34 tell people like code your MVP go make a prototype that's great but ultimately 8:38 you are going to want to have a developer who you trust who can follow 8:43 your vision and who you can hand things off to because that is their expertise 8:47 and they can do it about 1 million times faster now. And you are needed to run 8:52 the strategic like vision of the business cuz the product itself is only 8:57 one small sliver of what it takes to create a successful venture. 9:02 >> You're absolutely right. I mean selfishly I would love to just build and 9:05 code all the time but unfortunately you're absolutely right. There's a lot 9:08 of other business functions that supersede that and most of my role you 9:13 know I do leverage these these coding tools and most of the time it is to 9:17 create the demo to be able to show hey this is how I want it to look this is 9:22 how I want it to flow make it all work and it's interesting I didn't really 9:25 realize until getting into the process things break all the time there's bugs 9:30 everywhere you introduce a new line of code the whole thing falls apart it's 9:33 the most mind-boggling thing and you absolutely need somebody that knows what 9:37 they're doing to really help orchestrate that, especially in healthcare, because 9:41 if you've got a clinic that's running it and it breaks, like you need somebody 9:45 that's right there that can fix it, that knows what's going on, that's 9:48 non-negotiable. >> Yes, I I agree. You're right that like
Communicating your vision to developers and investors 9:52 getting your vision into some sort of either a prototype or even talked about 9:56 this on another podcast episode that you always hear about companies that started 9:59 with a drawing on the back of a napkin. You have to be able to communicate in 10:03 some way what is the thing that you're building to the people that can help you 10:07 build it. But then you need to also go convince the people that you want to 10:12 help fund it or the the people that you want to recruit to come be on the team 10:18 as your attention is drawn away from maybe sitting at your computer and vibe 10:22 coding all day. One of the other functions we talked about right before 10:25 the call was really how you create a runway for yourself and through 10:30 fundraising and uh whether it's angel investments or more traditional VC 10:34 funding, but how you can actually align stakeholders and how you approach even 10:41 just your capital table strategy because these are people that are going to be on 10:46 your cap table. They're going to be co-owners. They're going to be partial 10:48 owners of your company. So, I'd love to hear how you philosophically approached 10:54 fundraising and I love hearing about this whole investor advisor role that is 10:59 helping your company a lot. >> Yeah, absolutely. I don't have an MBA. I 11:03 don't have any business background and I use Chad GPT and Claude to help me. You 11:07 know, I didn't know what a cap table was 6 months ago or safe agreements or 11:11 anything like that. To be able to look at that, understand that my approach was 11:15 this is built for physicians by physicians. I'm very much a believer 11:19 that we understand the pain points, the workflows. We don't need another chat or 11:24 widget or button to click. We've all experienced various EMRs where these 11:29 solutions to help us actually end up hindering us. I ended up approaching a 11:33 very large alumni group with the Hawk Society. I had been giving these 11:36 lectures during our annual events and there was a lot of interest and we had 11:40 been talking about these solutions already and I I had a lot of interest to 11:45 build something like that. And so when we built it, I knew I needed some 11:48 capital to be able to really make this a reality. This is not something you can 11:51 necessarily bootstrap unless you have very deep pockets. So I approached our 11:55 physicians a part of our alumni network and said, "Hey, you know, this is what 11:58 we're building." We had nothing. It was literally an idea and we started our 12:02 seed our preed funding with safe agreements as a Ccorporation. I told 12:07 them, "Hey, here's your safe agreement." There's a lot of trust absolutely in 12:11 that stage, but they all believed in me. They believed in what we had done with 12:14 Morning Rounds coffee that we had proven some success in other places. They were 12:18 all in and even a modest amount, you know, a couple hundred,000 was able to 12:22 really set the ball in motion. Then we were able to get the MVP up and now we 12:26 have something to actually show something tangible and then that helped 12:30 flame the fire to be able to get others to come in. And what's unique about it 12:34 is as a physician investor, we've asked everybody to be a beta tester. as a beta 12:40 tester, they get first use of the product. And not only are they testing 12:44 it, but they're getting initial ROI just by using it and using the functions, but 12:48 they can orchestrate and be part of this physician advisory board to say, "Hey, 12:52 Justin, this isn't working. I don't like this. Change it or this is broken or oh, 12:56 this would be really nice to have. This is absolutely the top feature." And you 13:00 know, we were on with Todd recently. you know, Todd had a patient that that was 13:04 trying to refer through our system, but they didn't have the patient next to him 13:07 and they weren't able to schedule the patient. So, they wanted a link sent to 13:10 the patient to be able to refer them. We were able to build it that evening and 13:14 deploy it the next morning. And so, we've built this architecture now that 13:18 we can deploy very rapidly with very good and thoughtful feedback from our 13:22 physicians, which is super cool. Well, you think about starting a software and
Turning physician investors into aligned beta testers 13:24 you talked about all the bugs and you talked about how like things are like 13:27 definitely not going to work sometimes and how great it is that like the people 13:31 that are going to be giving you feedback are people who are like fundamentally 13:35 aligned with your success, right? They're on your team. They're excited 13:38 about it. They want to use it. They want to this company to grow and be very 13:42 successful and for their investment to be successful as opposed to like early 13:46 users who are beta testers that are just like, man, this sucks. Like toss it 13:52 away. Yeah, >> you're absolutely right. And you have to 13:54 frame it that way. You know, even with our betas, we say, "Hey, you know, this 13:57 is a beta. There will be bugs. I want to know about it. Tell me." You know, 14:01 because as an orthopedic surgeon or as a clinician testing software, you expect 14:05 it to work instantly. Unless you're like in the weeds, as you know, things there 14:09 are little, you know, this button doesn't click or this goes in a weird 14:12 place, you know, different things like that. And you can beat it up as much as 14:15 you want with your your QA team, but until you pressure test it, do you 14:20 really know what's going to work and what isn't working? 14:23 >> For sure. I think the other thing it does, the way you approach this, which I 14:26 I think is just so smart, is it puts people around you. I think it can be 14:31 very lonely to be a soloreneur, even if you're really driven by your vision. One 14:37 of the pieces of advice that I've I've heard from a lot of people is whatever 14:41 your vision or your idea is, like the first thing you should recruit is 14:45 adviserss. Even if it means like giving, you know, a totally reasonable amount of 14:50 equity of something that isn't worth anything yet. Like something that isn't 14:54 worth it. Like I literally have like a piece of paper. I I mean I did this 14:58 today on a new concept. I recruited a new adviser on an idea on a piece of 15:02 paper. It's not worth anything. But having some people in my corner as I'm 15:08 just getting started with the validation and the discovery on on different 15:11 concepts, it's like man, it feels so much more inertia, so much more like 15:16 motivation. You feel like you have a sounding board. You have a network to 15:22 rely on. It's like, oh well, I have no idea what to do about that, but I know 15:26 who I can ask. You know, >> absolutely. We've been very fortunate to 15:30 to have each other co-fellows, but we brought in one of our mentors, Mike 15:33 Mily, who is also part of our fellowship. He's business background, 15:37 Wharton Business School, and he's been really able to help be that sounding 15:42 board and a you know, this is how you talk. There's definitely a lingo and 15:46 terminology that you have to learn. Uh it's completely foreign and unrelated. 15:49 You mentioned this on one of your prior podcasts. We physicians love learning. I 15:53 went to school for 14 years after high school. It's it's a passion to get into 15:58 a new field, learn something, use your expertise. And our expertise is very 16:02 very helpful, but it's especially when you step into a different field like 16:06 business, you got to jump in, know your limitations, get the right people around 16:10 you. That's really been a huge part of our success is that
The changing SaaS landscape 16:14 >> there's another thing I was thinking about when we were talking. I've talked 16:18 to so many physician founders in the past couple months because of the 16:21 program we're running. There's a lot of people building SAS models. And so for 16:25 those listening that aren't in the software world, this is the software as 16:29 a service. This is like a QuickBooks or other services that you buy the software 16:34 platform and you subscribe to it. I think there's going to be some 16:38 challenges in that field with how much the market's become. It's open. I mean, 16:43 anyone can code now and everyone can code a lot faster now. What do you think 16:48 are some of the challenges in the SAS market and how are you thinking about 16:53 that considering that you are building a platform? like how are you positioning 16:57 yourself so that you're not going to be have a problem where I have talked to 17:01 founders where they're building something so narrow that I'm like 17:05 literally at some point every hospital is going to have a chief AI officer and 17:09 they're going to like just approve requests to just pre-made like just 17:14 custom make anything that people within the system want. So like your feature 17:18 could be something that Epic just turns on overnight and you don't have a 17:22 company anymore. So how do you think about that? building a SAS platform. 17:26 >> Yeah, that's number one question I get asked by my investors is what stops me 17:29 from, you know, copying everything you're doing. It's multiffactorial. To 17:33 speak to the the giants, it's easier to turn a speedboat than it is a, you know, 17:37 a cruise ship. And so, fortunately, you know, being a startup, you're very 17:40 nimble. You can pivot, but you have to have your eyes open to the future. You 17:44 know, in the last three months, we've seen more innovation in artificial 17:47 intelligence than in the last three years. We're on a rocket ship. I mean it 17:50 is unprecedented especially with a recent release of Fable Mythos and then 17:54 now Chad GPT 5.6 six saw. It's unbelievable. I mean, these tools are 17:59 awesome, but they're very scary. I would say the moat that we're building around 18:03 us is the security compliance, right? That's one. So, it's the infrastructure, 18:08 the company, the security compliance because that was not cheap. You know, 18:11 that cost us, you know, tens of thousands of dollars to to build that 18:14 and get that going. When we talk about artificial intelligence and, you know, 18:18 how does AI work? AI is a machine. It's a machine that's good at guessing at its 18:22 core. And it's only as smart as the information that you give it, the 18:25 context and the instructions of what to do with that information, the prompt. So 18:29 prompt engineering and context engineering very much have a huge
Prompt engineering, context engineering, and AI agents 18:32 influence on the validity and the the output that comes out of it. And so when 18:38 we talk about agentic systems, an agent is essentially an LLM, a large language 18:43 model, an AI like chat GBT and so forth that has a very specific prompt, 18:47 instructions, a very specific context, knowledge of what it's doing, and then 18:52 an action and tools at its disposal. And so where we're heading is we're heading 18:56 into agentic systems where you've got sub aents because AI is only good for 19:01 very singular finite tasks. When you start doing multi-layered tasks, it 19:05 starts to get dumb and you can't go over a certain, you know, limit in its 19:09 context window. And so you build sub agents orchestrated by super agents over 19:13 the top. The special sauce is not only how you organize that, but two, the 19:17 context engineering and the prompt engineering. If you look at my computer, 19:20 I've got Obsidian with a brain that's got all the files and everything. It's 19:24 collecting that, creating that table of contents so it knows what information to 19:28 pull when because you could dump all that knowledge, but you'll blow up the 19:31 context window and it'll start hallucinating. Very thoughtfully 19:34 crafting these prompts so that it's going to be replicable every single time 19:39 because especially in healthcare, it's got to work the first time and it's got 19:42 to be consistent every single time. We're making big decisions. And then 19:46 another piece of it too is FDA regulation, right? When we talk about 19:50 compliance with the FDA, we're very much heading into the days where it's going 19:54 to become considered standard of care to use these tools that not using them may 19:59 be considered malpractice and we definitely need to be a part of the 20:02 conversation in the development of this and how the FDA regulates it. But I'm 20:06 very much as we said a believer that there needs to be a human in the loop. I 20:10 think these are tools not to replace but to augment and enhance us to help call 20:15 attention to different things when they should be. But at the end of the day, we 20:18 are the decision makers and I don't think anybody wants to take over that 20:21 medical legal responsibility. >> This is such a great example of how you 20:26 as a physician, as a subject matter expert, as someone who's built and run a 20:30 practice, why you are the right person to build something like this. A lot of 20:34 people, especially doctors, don't realize how many companies are built by 20:40 people because they had an experience in that realm or they like had a a family 20:46 member who had that issue. The best example is some endocrinologist I was 20:50 working with. She was like, "This diabetes company called me and they 20:53 don't have anyone clinical." And I'm like, I know there's people building 20:58 solutions for really good reasons, but they don't actually talk about context 21:02 window for the agents. Like context window as a founder, like you are 21:06 actually bringing your context window as a surgeon, as a physician, as someone 21:11 who started a practice, you could be very very strategic about how you bring 21:15 that subject matter expertise to to build something that other people 21:18 cannot. >> You're absolutely right. It's because I
Using AI internally to build a leaner company 21:21 speak both languages that this has so much power, right? I've had several 21:25 medtech companies call me and they're like, "Yeah, we'll meet up like 21:29 quarterly." I'm like, "Qorderly." You're going to have zero insight in what 21:32 you're building. It has to have a lot of organic and very consistent, especially 21:36 with the pace that things are moving. The company is building AI, but we're 21:40 using AI to build AI. So very much throughout the company, we're using 21:44 artificial intelligence to help orchestrate and not only keep our burn 21:48 rate low, but be able to be nimble and keep things moving as quickly as 21:52 possible. Because if we're building a platform to reduce overhead for 21:55 physicians, why aren't we using it or a similar type concept internally to make 21:59 sure that our business is running as lean as possible too? 22:02 >> It's a great time to be building. I would love as we wrap up if you can 22:05 share when this is out in the world and successful and at scale like what's your 22:10 vision? And how is medicine better with what you're building? 22:13 >> So what we're building is an agentic solution to your clinic. It's specialty 22:18 agnostic and essentially we're working through the food chain through 22:21 automating intake triage patient debrief as if a fellow was in the room coming 22:27 out and presenting everything to you. obviously AI scribe and then 22:30 synthesizing that into your medical legal document, referring provider 22:33 document, your patient education note and then billing and coding and just 22:37 moving through the food chain and trying to make it so that there are as little 22:42 clicks as possible for a physician. The beauty with AI is now we have the 22:46 ability to take so much information and synthesize it whereas before it was just 22:50 daunting and took too much time. So being able to pull in wearables, being 22:53 able to pull in other data, synthesizing it so that now we get actionable things 22:58 to help our patients peroperatively, nutrition, all these different things 23:02 where it was just too much time or there's not a CPT code for it, so I'm 23:06 not going to, you know, I've got 15minute consults and I've got eight 23:10 patients every hour. So it's helping to deburden that and just give physicians 23:14 that opportunity to spend more time with their patients, build that relationship 23:17 which AI will never take over and then two be able to come home at a reasonable 23:22 hour and have time with your family and your kids and not have to be doing notes 23:26 or administrative work for two hours need.
Making independent practice more accessible 23:28 >> I think it also lends a pathway for people to start a practice who otherwise 23:32 might not have started a practice. Right. 23:34 >> Absolutely. I was building it for myself and that was part of the name with some 23:38 of our partners. We're not stopping just at the ambulatory practice, but building 23:42 layers on top of it to help be able to run the business operations of a 23:46 practice. The future is bright and we're really excited. 23:48 >> That's great. As we wrap up, how can people find you, Justin? 23:51 >> Yes. So, if you want to reach out to me personally, you can find me on LinkedIn, 23:55 Justin T. Smith MD, and then our website, if you would like to submit, we 23:59 are still taking physician investors that want to be betas as well as early 24:03 users. Uh, we're aiming to go to market here in the next several months. and our 24:07 website is turnkeyipractice.io. >> Great. Well, thank you so much for 24:13 sharing your journey and your learnings along the way. Um, I hope this was 24:17 inspiring to some of the people listening who also have ideas of things 24:21 they want to create. The best first step is to take the first step. So, 24:25 definitely feel free to reach out on LinkedIn to Justin or me. We love 24:30 supporting other physicians on their journeys. So, thank you for joining me, 24:34 Justin, and thanks everyone for listening. 24:37 Yeah, thanks for having me, Alison. >> Thanks for listening to the Startup 24:40 Physicians podcast. This community was built for doctors who want to contribute 24:44 to the future of healthcare. We offer support for physicians who are ready to 24:48 make the next step towards building the career they've dreamed of. Visit 24:53 startupphysicians.com to learn more. And if this episode 24:58 resonated with you, share it with another friend who needs to hear it. 25:01 Let's change the future of healthcare together.
Comments