BoneChat #54 - Discussion of AI tools for orthopedic businesses

So the call I was just on, um, was with a couple of big data scientists from Utah, and they're both UC Berkeley people. And I told them what I was doing with, you know, with generative AI and the AI agent, and, uh, you know, how we're basically, you know, using it as much as possible and learning more about it. And I suggested that they write a book called, you know, AI for the Complete Idiot, uh, or something like that, for most of us.

AI for Dummies.

Yeah. They, um, one of the things that they, they're, they're really, they're talking about is that we're just, we're kind of on the edge of an AI disaster. And a problem that they see especially in healthcare, is the relative limitation of our data. Almost all of our healthcare data comes from places like Harvard, Boston. There's just a handful, less than 6 urban areas. And so we don't really have data for rural people, so to speak. And we've kind of run out of that. Um, we kind of run out of what we have. And then they start talking about the effects it's having on some of the, uh, some of the medical specialties. Like in Florida, just passed a law that the receptionist can take x-rays now. Wow. You know, you don't have to be an x-ray tech, uh, except maybe for just the heart or something, or the brain, uh, which is kind of depressing. Uh, you know, radiologists are quitting in droves. The average retirement age for radiologists, 10 years lower than it used to be.

Well, it's all remote. Radiology is remote now. It's all done in Pakistan and India and stuff, so they can't compete.

Uh, so I mean, it's just kind of, it's just kind of an interesting discussion, so to speak. They're trying to, they're trying to publish a book, and they're trying to figure out whether they want to call it a, you know, just a warning or just an analysis or whatever. And I, you know, I told them, I said, you know, hey, sex, fear, and greed always sells. You know, put a, put a naked woman on the COVID you'll sell it.

Who knows? Healthcare is so, you know, healthcare is so bureaucratic, you know, it's, and I, I wrote an article about this years ago. I think there's, I said in 2025, there's gonna be a gold rush for data. And I could, and, you know, having millions of patients in a central database is gonna be worth gold. But I don't think it's doable. Like you said, it's there, every, you know, in the big cities, everybody's got their proprietary clinical study data. I remember Zimmer, I had a meeting with Zimmer like a long time ago and they had something like, oh, what did they have? They had 10,000 CAT scans of knees. And, but they never did anything with it really. And now maybe that, maybe that's valuable now with AI.

Maybe they can, yeah, it does have a value. Yeah. One of the guys on the call, his name is Edward. They're, they're, they're both Russian Jews and they're really interesting people. They, Edward was saying that a, a good study now, an imaging study has a value between $1,000 and $3,000. Kind of a street value. Who knows? I mean, you have to find a buyer, I suppose. Yeah.

So, so thanks everybody for making it because I kind of botched the link. I don't know how you found us, Elizabeth, but so I scheduled the first 50 webinars and that you can't schedule more than 50. And so now I had to make a new lot of another 50. And so I think it's gonna be a different link, so I will fix that for next time. But how'd you get in, Elizabeth?

You sent me an email.

Perfect. It worked.

Okay.

That second— yeah, the new— the link in the email works then.

Perfect.

Perfect. So our discussion today, Mark was gonna have a— we were gonna do 2 for 1 AI tools and orthopedics, but the other guy dropped out with the conflict. So we thought we'd just have a general discussion about, you know, what AI tools has everybody seen, what works, what doesn't work, you know, how much data do you have to have. I mean, we can, I can, I can share my screen just for, just for thought-provoking here. Uh, let's see if I can do this. I'm going to fire myself as an administrator. That's not it. That's not it. I can't find it. Shoot.

Also, the guy that couldn't make it, uh, his name is Adam Hocek. Um, he sent me a video of what they do, so I can show that video. It's only 3 minutes long or so. And the reason he, he couldn't make it was because he had the opportunity to do an investor call Oh wow, that's understandable. So he blew us off for an investor call.

Yeah, he's not going to get invited back, that's for sure.

Yeah, I told him, I said, trust me, out of all the things you could have blown us off for, that's one that everybody will understand.

Yeah, it's so hard to get in front of investors now. Okay, I'm going to try to— man, I'm so bad at this.

So Tiger, if, if, if Not many more people show up. Do, do you want to push this, uh, conversation?

We can talk for 20 minutes, 30 minutes, whatever you want to do. But I—

this—

while you're doing that, let me show the video.

Oh yeah, okay.

Uh, yeah, let me share my screen.

Far away.

Hang on, let's see here. Let me pull up It's on YouTube here, so I'm always— I always tell people, said, when I pull up my YouTube to show video, don't judge me, okay? Here we go.

On your suggestions there, or your SNL sketch?

Yeah.

Okay.

Welcome to Echo Recruiter, our AI-powered talent acquisition platform. Using advanced speech pattern analysis, we provide comprehensive candidate assessment in under 60 seconds. Our language-independent technology analyzes over 20 key dimensions, delivering unbiased insights to make the best recruitment decisions. In this short video, you'll see how we're revolutionizing the hiring process with cutting-edge AI technology.

Wow.

Our job creation module streamlines the hiring process. Recruiters can quickly define role requirements, skills, and competencies. The platform aligns requirements to ensure consistent and objective candidate evaluation. This approach eliminates traditional hiring biases while maintaining high assessment standards. Let's see how candidates experience Echo Recruiter. After receiving their interview link via email, candidates begin their journey with a single click. They're welcomed into our streamlined application process where they quickly verify their email and provide basic information. The final step before the interview is uploading their resume. Our AI immediately processes this to personalize the upcoming conversation. With everything ready, candidates can now proceed directly to their AI interview. Now, let's witness our AI interviewer in action. Using advanced speech pattern analysis, our AI conducts natural, adaptive conversations that assess both technical skills and personality traits, all while eliminating traditional biases. Greetings, John. I am Emma, your AI interview assistant. In this session, I will guide you through your interview process. To get started, tell me why you think you're a good fit for this role.

I believe I'm a strong fit for this business analyst role because of my experience in data analysis and my ability to interpret trends to support business decisions. Confidence.

Thank you. Can you give me an example of a time you solved a challenging problem in your previous role?

Certainly. In my last position, we faced an unexpected drop in user engagement. I conducted an in-depth analysis, identified patterns, and recommended a strategy that improved engagement by 30% within a quarter.

Thank you, John. This concludes our interview. The recruitment team will be in touch with you soon, and you will receive your personalized feedback via email.

Pretty cool.

It's got a little bit more, I think.

The recruiter dashboard provides real-time visibility into your hiring pipeline. Track candidate progress manage interviews, and access instant analytics.

Turn it off.

This centralized command center streamlines the entire recruitment process, allowing teams to make data-driven decisions efficiently.

You know, that, that's a really cool tool. It's like a— I mean, Elizabeth's like a sociologist, and Mark almost was a chaplain, so— and I'm in the people business, so, uh, I mean, the— what I see, it's— could be really helpful to HR people, internal HR people. But what I see too much, having hired hundreds of people and placed hundreds of people, the companies don't know what they want. I mean, it's a— it blows my mind, you know. I need a director of marketing that has these, you know, skill sets. Okay, here's, here's 5 to 10 candidates. Well, that's not exactly what I want, you know. You know, number 8 looks interesting, but they're missing. And we always have to recalibrate. It's incredible how often companies don't know what they want. And then they end up— I mean, I'm just, I'm crying on the shoulder of other recruiters here, but they end up hiring people that they are comfortable with and they're like, and they're more like them. They don't hire based on the criteria that they originally put out. It's really interesting.

It goes to talks. It's basically like how they feel with the person. Like you said, are they comfortable with them? Do, um, they check all the other boxes, but if they are comfortable with them, you know, thinking they've got to spend 8, 9 hours a day with this person, they want to think flow easily.

There's, there's an airport litmus test. Can I see myself sit in an airport for 6 hours with this person? I've heard that before. Yeah, it is. It's really interesting. Yeah. And I tell candidates that, you know, when you go to an in-person interview, you're already qualified. I mean, hands down, you don't have to prove that you know how to do the thing. It's all about fit. So the CEO or whoever's on the other side of the desk is going to be saying, Do I see myself working with this person every day? Exactly what Elizabeth said. And that's, that's the real, that's the final question.

Anyway, I just put in the chat, someone, this is November 18th. Article on Fortune magazine that gives you a little bit of an idea of what they think the impact of AI will have in the healthcare market. And when you, when you say it's going to grow from $19.5 to $27.6 billion, that's a big jump. And so I know it's going to have a huge impact. So, so Tiger, my, my interest in AI is not from a geeky kind of a technical perspective. It more stems from the fact that in my regulatory compliance business in the last 5 years, I've tried everything but basically starting a war to market a pretty boring product to medical practices and hospitals and just try to speed up the growth. So my, my, my lead generation budget has gone to several different companies. Everybody promises the moon. Everybody says they'll get the message out, they'll do the branding, they'll do this, they'll do that. And basically, in almost every case they can't get to the decision makers in our space, and they're just too hard to reach. You know, doctors, uh, physicians, you know, very often make more money than they can spend. They just want to get home and have dinner, uh, you know, whatever. Hospital administrators get paid to go to meetings, and they don't really need to move the needle, it seems like, unless they're at a certain position. Uh, plus compliance is not something that makes money. And so I've, you know, so I've tried all these lead gen things and so forth, and I'm always looking for different ways. And I just think, you know, my gut says that at least AI will affordably and economically enable me to, during maybe a period of, of sort of new interest, uh, help me a little bit. So that's kind of why I've been diving into it a lot. Plus, I have an AI model that I intend to market to the entire healthcare industry pretty soon, and And I could just real quick show you one, one more thing how that works. Let me share my screen and give you an example of that. And if I'm able to, um, I do— let's see, you ever heard of the company Abanza?

Yeah.

Yep. Spanish company.

They got a better ACL fixation.

Yeah, they got new ACL things. So I, this morning, I put together just a really quick AI model for them. And so the, you know, here's a phone number that somebody could call. So I could either put this phone number on my website, or I could put it into a text or an email, you know, say, hey, Dr. Smith said for you to listen to this, please do. And, you know, if the guy trusts the, the click, you know, the, which nobody does anymore, But if you can get them to click this, then they're going to listen to this message. Here we go. Let me call this number and you'll hear it. It'll be live.

Hi, I'm an AI representative for Abanza. I'm reaching out today to talk about a solution that can improve your soft tissue fixation outcomes, particularly for ACL repairs. I'd like to show you how our WasherCap technology can make a significant difference in your procedures. The WasherCap provides a more secure fixation, reducing the risk of complications and revisions. It also allows for a more efficient procedure, saving you time in the OR. Additionally, it provides a lower profile, which is beneficial for patient comfort and reduces the risk of tissue irritation. I'd love to show you the WasherCap in action and answer any questions you may have. Would you be available for a live demo next week? What's your schedule looking like?

Um, maybe Tuesday at 5 or Thursday at noon.

I think Tuesday at 5:00 AM or Thursday at noon work great for me. Let me just send you an invitation for the demo so you can add it to your calendar. Which one would you prefer, Tuesday or Thursday?

Thursday.

Got you down for Thursday at noon. I'll send you a meeting invite shortly. Just to confirm, we'll be discussing the washer cap technology and how it can improve your ACL repair procedures. Do you have any other questions for me before we wrap up today?

Do you need my contact or email? I guess you don't.

It sounds a little bawdy, you know, I just don't see surgeons, you know, well, let me share, let me, let me stop my sharing here.

Yeah.

So I mean, I mean, it's kind of interesting, but it's kind of early stage and I don't really know yet whether it's gonna be effective or how it's gonna be effective, whether you could just put it on your website or whatever, but it's just the beginning.

Yeah.

Yeah.

It's early days. Yeah. It's just, I think there's gonna be 1,000 niches in healthcare where people come up with little AI agents and I don't know what those are yet, uh, but it's— they're just going to be everywhere and you're not going to be able to know it. Um, you know, it's got to, it's got to sound like a real person. There can't be a time lag, obviously. You know, that, that'll go away in the next year, probably. Uh, I mean, here, let me, let me share my screen and we can— everybody sure is quiet. See if this is working. There we go. Can everybody see that?

Yes.

Yeah. So the yellow, the yellow are kind of business areas and the numbers are kind of questions about can AI help you in that area? So general business, marketing and sales, Mark talked about that a little bit. Customer service. And like Maurice— Mauricio talked about that last week or two. You know, you— it's great for inbound. It may— there may be some opportunities there. Finance and accounting, logistics, decision-making. I mean, there's just so many areas. You got to figure out where's your pain. I mean, one of the questions is, what do you do that's a repetitive task? Um, and then what do you do that is a complicated data task? So like reading contracts, reading spreadsheets and trying to find trends, reading clinical data and trying to figure out if things reach endpoints or not. I mean, there's, and there's a tool for all these things. There's, I just, uh, my Mark mentioned last week that he doesn't use attorneys anymore, and I just did a quick AI search, there's like, there's like 20 AI legal assistants now. I mean, just, I mean, it's hard to even choose them, there's so many.

Um, I mean, I use attorneys for basically like a doctor uses his EMR. When he's using an AI assistant, he reads what they wrote and makes sure that it's accurate. And that's how I use my attorney. Of course, you better have a good friend who's an attorney because you're kind of screwing them over a little bit, right?

Right. It's a final, it's a final stamp of approval, but they didn't do the work.

Yeah, no.

And so I use it for writing. I mean, you know, I write every day and, you know, it's speeding up my writing because I'll have an idea for— like, I just wrote something called, you know, you only need 25 customers to, to have a successful business. I wrote that last week. And now, I mean, get, get this, this will blow your mind, Elizabeth. Um, I'll write a draft. I'll just say, okay, I want to write a blog article about this, this, this, and this, and I kind of make an outline. And then I say, and put it in the voice of Tiger Buford at Orthostreams. Wow. And it, it, and it sounds, I mean, in writing, it sounds like me. You should try that, Elizabeth. You should, you should write, you should write a draft article and say, put it in the voice of Elizabeth Hoffheinz, and you'll make it multiple personality disorder, and I'll confuse it.

What, um, okay, so which, which program was that exactly?

It doesn't matter. They all work.

I use, I I use Grok now exclusively. Grok is the— it's on Twitter. It's— and Grok has no biases and it's super fast. And, and I think Twitter is going to pass OpenAI probably. They're, they're moving so fast. Interesting. Yeah, Grok. But ChatGPT, you know, if you, if you use the, uh, the O1 version, you know, the, the paid version, it's really, really good.

Okay, according to the, the, the UC Berkeley scientist guys I was talking to a while ago, that they all have deep biases. And because all of the data tends to come from, uh, in their— in his words, and like I said, he's a Russian Jew— he said they all come from educated white people. And of course, and he speaks very plainly because, as you can imagine, you know, this guy, he's an immigrant. But I love talking to him because he just really, you know, talks so pointedly, I guess you could say. There's no nuance to his—

Yeah, direct, no filter, direct.

No, no filter, not at all.

So I mean, I use AI to read contracts. I mean, I, you know, I deal with non-compete agreements and recruiting contracts and NDAs all the time, and I just literally just throw it in there and say, tell me what this means, like I'm an 8th grader. And, and it's like, boom, okay, now I just saved myself 30 minutes. He's just like, I mean, it's just incredible that, um, it's— so there's kind of two things we're talking about. One is what can we use as personal tools for ourselves or our business, and the other thing is what Mark's talking about is what business opportunities are out there to solve some problems in healthcare.

So, I mean, I have learned with this AI engine that I'm currently working with, with these Canadians, that whatever knowledge base that I feed into it, they can then speak knowledgeably about it. And so I haven't had the opportunity yet, but if I could feed into it everything about your medical practice, what your specialties and subspecialties are, what your hours of operation are, what services you're billing, information. The more, you know, information I put into the knowledge base, then that AI could become now the receptionist that takes all the calls or the after-hours calls. Uh, there's a guy I know named Ravi Rahija. He's a a pediatrician by origin training, but he and his PhD wife started a business called Triage Logic. And what they do is they take the after-hours calls for primary care practices and internal medicine practices when people call about, you know, should I take a 5 milligram or a 10 milligram, or, you know, whatever kind of after-hours stuff. You know, that it might be, so they don't get bothered in the middle of the night by somebody's ear infection. And so, and you know, he's hugely interested, and he's kind of wondering— he has 150 RNs that take these calls, and he's kind of beginning to wonder how much of this could be done by these AIs. Is he going to put these RNs out of work?

Well, he needs to feed the system with like a ton of calls, and then he can just get rid of the RNs.

That's— I don't know. And I was on a call the other day where one of the people on the call was a lawyer, former RN lawyer, who has put together an ethical oath for healthcare and AI.

Smart.

That's all unknown territory.

Wow.

I don't know.

Things are changing fast. It's just, it's changing way faster than the healthcare system can handle it. That's the problem.

I—

yeah.

What do you think, Beth? You're a very creative woman as a writer.

Yeah, it's not going to get ahead of me. I'm determined. I think, um, to— you know, the nuances of being a human being cannot fully be captured by, uh, AI, and that's probably what will save us.

Right, but there's repetitive things that you don't want to do, like you don't want to read contracts and— yeah, you know, it's, it'll, it's a— I just see it as a time saver and a, it's almost you can do more with less employees, you know, it's— but it's just got to figure out all the tools. Eric and Kyle, what are your thoughts?

I mean I have, I have questions. I don't have personal thoughts because I don't know enough yet. So what I've been wondering about with the, the, the, the interview tool that, that, that you showed, then Tiger, with what you were talking about with, you know, reading contracts and summarizing it for you, how is, is the, the output from that validated, like, like with the—

It's not, it's the average. So it has, so these, these large language models have scrubbed the internet for years and they, they have a general knowledge of everything on the internet. And so they give you the plain vanilla down the middle response of the collective knowledge of everything that's ever been on the internet. And it's not validated. You have to, like Mark said, you'd have to go to your attorney and get somebody to validate it for you. But it gets you like 90% there, you know.

Well, with that, with that interview tool, you, you know, I wouldn't want to use it until I know that it was giving me good results. So that's why I was wondering, how do you, how do you know that?

Good question. I don't think you could ever validate results in recruiting and being in the business. It is so human and so messy, and it is— people are irrational and they, they write terrible job descriptions and it is so messy.

Well, you know, like what Elizabeth was saying, I, you know, I, you know, being at Medtronic for so long, they had these horrible interview questions that everybody had to ask. And I refused, and they, you know, never fired me. But I refused to ask any interview questions because it didn't matter. I mean, to me, they were canned responses anyways. It was more, you know, I had a conversation with people, and that was much more important to me.

And it makes me— what you're saying makes me think of actually something that I can't imagine AI would be able to help with. A doctor told me years ago how he was selecting residents for his program. He said, take every one of them, all the last ones standing, to golf and see how they handled disappointment, frustration, you know, joy, whatever it was. And I thought, oh, that's pretty interesting. You know, it's live in the— oh, I'm being summoned.

I will be back.

Pardon me. So, so the other thing that I've had trouble with, with, with AI, and it's my lack of knowledge. I don't know what all it can do for me.

Well, start— don't start there. Start with the problem. What problem do you have? What takes a lot of your time? What's hard to figure out? When do you need to hire an expert, you know, that's expensive?

Let me give— let me sneak in here with a quick example of something that I tried a few months ago. I haven't put it into practice yet, but was rather blown away by— not so much blown away. I was pleased with what I saw. So if you were to go to any of these large language models and, and you hear that terminology thrown around, Eric, essentially to summarize what their algorithm does is it looks at a series of words that you type in for a prompt and then it goes into the internet and finds out what is the next most likely word in the history of the internet to happen after all those words are put together.

Yes.

And then now that that word has been found, what's the next most likely word after that? So it essentially takes an accumulation of repetitive or familiarity with words and then just puts them in a string together. Now, if you then work with it to refine itself, it can then start to learn— and that's not even the right way to say it— but it can then start to hone in on what you were looking for. So what I tried a few months ago that was rather neat was I said, all right, we're going to develop some design control documents for a startup company. Following ISO 13485 and 21 CFR Part 820. Please reference those specifications and write me a sample quality plan for a trauma med device company. And it thinks about it for 3 or 4 minutes and it spits out something. And it's a basic template that's got a header and a footer and a title and a purpose and objectives and references and responsibilities. And then you read it and 30% of it is crap, but 70% of it is really good. And the structure and the organization and the format and the tone of it is really directionally on point. And my total cumulative amount of effort to get to version 0 of a document was 5 minutes and I typed 60 words. So that type of thing I think is really going to be fascinating to see how that works because I then I then kind of carried on down that cycle and said, all right, write me a risk analysis plan, write me an IOVV matrix. And it, it just pulls from stuff that's out there. So I think some of that is very interesting in the space that many of us work, where because we have regulations to draw from, being able to point an AI model to a regulation that then gives guidance to that model to give you an output that saves you having to to redo something. I think there's really something there. And then through a variety of elements of this conversation, you guys have talked about redundant work or repetitive work or assistant type of work. I had an experience this week with my veterinarian that would have absolutely been totally replaceable by AI. Now, the ladies that work in the office are wonderful. I hope they don't lose their job to robots, but absolutely could have been totally replaced by AI. I needed ear ointment for my dog. It's like the 19th time I've needed ear ointment for my dog. I needed his heartworm pills and I needed his itch medicine because he's a stupid Labrador. He's the best dog ever, but everything is wrong with him because he's a pure Brit. So I had to call in to speak to this receptionist and say, I need a refill. This person had to then take my information down, go into my file, look at the last time I had an appointment, type up an email or leave a note for the veterinarian to then say, do they want me to come into the office or can they just fulfill my prescription and then call me back and tell me that yes, you have it and bill me. Now, if you take that to the extreme of what is possible with technology in the world, I should have been able to simply leave a voicemail for an AI to take a look at. It would've been able to draw all the necessary patient data for my dog, send an email to the, or a text to that, that veterinarian. They don't even have to be on site or even care. They could say, Kyle called, he's calling for Cooper, he wants Apoquel. The last time they had it was August 17th. He also wants ear medicine. The last they had, we recommend that you approve this prescription. They could click yes, and then a drone could pick it up from their inventory and drop it off at my house, and they could charge my card on file. That's the world that we're moving towards is that sort of thing starting to happen. And it'll be fascinating to see how we apply all of that. So that was my little soapbox. I think it's an interesting place to start charging into.

So Kyle, you've identified something that is the crux of where we are, and that is literally if you can imagine an AI to do something you know, that's sort of, you know, task-oriented or mental and not physical. Of course, literally it can do it. And it may only be 80%, but it'll draft out, you know, the engine. If you could— we could right now put into an AI, right, I want to write a clinical study for this, this, and this, and it'll give you a draft that someone may take otherwise 3 days to write, and then sort of go from there. It's pretty interesting. I don't want to take the time of it, but I mean, I could right now construct an AI, you know, person and feed into it, say, you know, limited knowledge base on what either one of y'all, you know, might be doing. And it would give me a pretty good persona to make a, you know, an overture or a phone call or answer a call or whatever. But what you just described about the veterinarian, that's what I was talking about earlier. If you chunk into my medical practice, you know, everything about the practice, you know, the hours of operation, our specialties, what we all, we all, we offer, the billing information, so forth and so on. And AI could handle that quite well, frankly.

Yeah. I'll give another practical example because this, I think it's a great topic. I love it. Think about, think about, so I work in the consulting world for engineering services now. If someone came to us tomorrow and said, hey, we need help. We want you guys to, to be the, the team that develops a dual mobility hip for us. One of the first tasks that I would do was I would look up all the IP that's out there, or I would look up and get, and I would get an understanding of all of the competition. I would look at all the materials. I'd start figuring out who my predicate was. Now, it should be a standard practice for me and for the engineers I work with, for the very first thing we do with those is to just throw those at AI. It's free. And yep, it won't do the job for us. We're going to have to come back and we're going to have to do that exercise. But we may be able to let computers do free work for us where we can take a couple of hours and they can spit out an output that actually represents and replaces 20 hours of just pure internet legwork and nothing is lost. Now, the job isn't probably done. No, man, it's, it's, uh, It's gonna be really cool.

So just this morning, from a little editorial from the New York Times on early morning audio, one of the reports I, I read that AI is having almost the biggest impact on small businesses, but with between 1 and 4 or 5 employees, you know, that generally a lot of us, you know, are in this category. You know, they run small consulting businesses or small businesses. And, you know, we're jumping on this because, you know, we're educated, literate people and we, we see the value. And also we don't have the backend necessarily, or, or we farm it out and don't wanna take 3 days to do it.

Yep. Mother is the, the invention. What, what's that saying? Mother's the root of all necessity.

Necessity is the mother of invention.

There you go. And little, you know, little companies like me and you guys, we don't have a lot of people. And so let me talk about prompt engineering is becoming the new skill. So if to use any of these AIs, you have to be able to ask the right question or set up the question correctly. And then you get the 80% correct answer. And then you ask another question and you keep massaging it. And prompt engineering. I'm trying to spend a lot of time learning prompt engineering. The way you ask the question can help you get the best answer. And it's a whole new skill set called prompt engineering. Just go to YouTube or anywhere. And that's, that's the hardest part. It's a new skill set. The AI can do anything, but you have to ask it correctly to get, to get what you want.

So I had my first foray into this just yesterday. So I'm a late adopter as always. But I had a presentation that I'd given. It was like 30 slides, you know, it was a biomechanical presentation and I want to give it at another meeting. This meeting requires an abstract, but I had the presentation done already, so I copy and pasted the PowerPoint slides into ChatGPT and it gave me an abstract. You know, I said 300 words max and stuff like that. It gave me a good start. I would, I would say it was, you know, 60% good. So I think that's, that's kind of what, what, what you all are saying. But again, you know, I probably didn't give it—

you gotta ask it, you gotta ask it over and over again. You know, refine this, tell me more about that. I want to see it from this viewpoint, not that viewpoint. You just have to, you have to have a— it's really, really good for brainstorming, but you have to have a conversation with the thing, you know.

And also realizing, you know, I have an IDE submission going in next week and just thinking about you know, all the hours I spent writing in each different section, how much help I could have gotten.

Here, let me share this with— I showed you guys I have a— I created a digital twin of myself, and this one came in. Come on. Here it is. So this one came in. Somebody, somebody came in and said, does it make sense to open a branch for Surginova in the United States? And this was my answer, my digital twin's answer right here. And the amazing thing, I mean, it's a pretty, it's a, you know, high-level answer. I don't know if you can read this or not, but the amazing thing is there's references in here. So there's a YouTube reference from Patrick. There's a, there's a YouTube reference, Anatomy of a Startup. And there's other podcasts. I mean, it's incredible. Uh, so, and then the, if the person kept asking questions over and over again, he probably would have gotten a better, a better answer. But, um, it just happens in the middle of the night, you know. I don't even— I, I wasn't around. It's, it's amazing.

Wow. How, how did you do that? This is, this is one of the things I don't— yeah, so I find this, and how do you—

yeah, so I'm all— I'm, I'm the opposite of you, Eric. I'm, I'm the cutting edge on everything. I'm looking I bought Bitcoin in 2014, you know, but I'm, I'm using— I, I was on the internet in 1995, you know, dial-up modems and everything. So, um, so there's a— I just keep reading and learning, but there's a, there's a company called Delphi, D-E-L-P-H-I, and you just go to delphi.com and you can create a digital clone, but you have— the hard part is you have to feed it And so you, you have to have a lot of videos of yourself or you have to have a lot of written or audio of yourself. And so I just fed it all the blogs I've ever written and all the YouTube, you know, things I've ever, ever done. And it, it, it takes about 24 hours to create your clone. And then it's like, it costs like $20 a month. And, um, yeah, that's, that's, and And then you end up with a website. This is— I don't know if you guys can see this or not. Crap, you can't see it. Why is this so bad? Well, let me try to—

What's the website?

Well, you end up with a— it's delphi.com backslash your name, you know, in front of your face. Yeah.

Or in front of your chest. There you go.

Right there. So this is the website. When you go to— you go to my clone, you go to this website, there's two buttons. You can talk to the— you can talk to it live, or you can write, and it gives you suggested questions. And, you know, it just— anyway, I'm, I'm always trying to figure out a way to clone myself. So Delphi is a good start, but it's, it's real, real early. I mean, Elson, I'll put it in here for you guys if you want to play with it. Yeah, so here's the— put it in the chat. So it's delphi.ai. I was wrong. Um, delphi.ai backslash whatever your name is, is, is— it creates the— that's the name of the clone, and it takes you to a website. And then pretty soon, I, I think within a year, I'm going to have a talking head. That looks and sounds like me. And I may even— I may even try to fake everybody out on a phone chat one day and see what happens.

Wow. So I'll keep you guys appraised of the— how the, um, the AI deal is doing for lead generation Customer service and for VenOps. I've got one for VenOps too, but I won't— you know, it's basically the same thing. It, it, it talks to you. It really knows a lot about VenOps, right?

The problem is what Mauricio said last week, is you, you don't weaponize it for outbound. You're going to just piss people off. You can, you can use it for inbound, because when, when a doctor reaches out to you they're already— there's a little bit of trust there already. And they, if they get a bot that, that helps answer the questions, they're okay with that. They've given you permission, basically.

I think so too. And that's, that's why I think initially this lead gen will be best served by just putting it on your website and say, you know, this is Charlie, he's an AI agent, he'll answer most of your questions. If you're interested, we'd love to talk to you further. And you click that. That way you know it's a safe click because I think it was literally just yesterday or today, uh, Forbes came out and said don't click anything, which is kind of where we are. And also said if you don't know where the URL came from, don't scan a URL.

I mean, yeah, it's, it's hard to trust anybody.

Get to the point where how do you communicate?

I think I really— everything I read says that in the, the next couple years, the government is going to make it a law that if it's a bot, you have to identify it as a bot, and if it's a real person, you don't have— that's, that's going to happen. Um, so you, you know, so it'll, it'll— you can't fool people. Um, same way there's a law with marketing emails. You, you can't, you can't force people to sign up for your newsletter. You, they have to opt in and you have to give them a way to opt out. Yeah, by law. But previously people would just spam, they just spam a million people with their, back in their good old days.

Well, Eric's comment about how do you validate this stuff is, is kind of intriguing and interesting. You know, how much can you trust it? How do you validate it? I know that the guy who does the Echo AI, he is, um, a psychologist. And his— and, you know, in my, my work, you know, at USC was in behavioral assessment. So I, you know, personality profiling is a science, and this guy has simply reduced it to a voice recording, and he's as accurate, he says, as any Myers-Briggs or any, any other typical assessment tool. I bet. Which is kind of interesting, really.

And then the key is it's very little information that's needed, just like you said, 30, 60 seconds of voice. And that's incredible. Here, I'll blow your mind, Eric. So, one of Elon Musk's companies is called xAI, and that's— he's creating his own AI company. You know, he's got lawsuits with OpenAI. But anyway, you may not know it. The data center is in Memphis by the river. He's built this massive data center and they're buying like a million, you know, Nvidia chips. They keep upgrading it every week and this will blow your mind. So FedEx is Memphis company, great company. They've been around forever. They're worth $75 billion market cap. XAI just started. They're worth $50 billion and they're in the same city as Memphis. I mean, they, they literally just started. And they got a valuation of $50 billion, they'll pass FedEx so fast they're not going to know what's going to happen. It's just, there's a gold rush for data and for AI. It's incredible.

I organized a dinner 2 years ago, uh, at the Academy, and I, one of my people I invited was Robin Young. Yeah. And I invited a lot of other people. Most of them were CEOs or, you know, or founders and. And somebody asked Robin, how do you make a lot of money in orthopedics these days? 'Cause they found it a little frustrating. And he said, well, he said, if you wanna make a small fortune in orthopedics, you start with a large one. That's, and he said, if you really wanna make a lot of money, he said, big data is the new oil field.

Yeah.

2 years ago he said that. And it's even more true now, I think.

Oh yeah, it's— and so the, the device companies, you know, can't do this stuff. You know, J&J and Smith Nephew, Zimmer, they can't, they can't do any of this stuff. Um, it's going to be the little guys that come up with new applications that are— that people are going to have to have. Um, yeah, I don't know where— I just don't know where everything's going. And it's, it's— everything's basically free. Right now too, you know. I mean, I just created a clone of myself for $20 a month. That's ridiculous. Yeah, you, I think you just gotta find where's the pain, you know, where's the pain in your business or your consulting, your service, your sales. And then you gotta, there's probably an AI tool that can help you with that pain. Yeah. Inventory. I mean, geez, that's a nasty one. Inventory management.

It was funny you mentioned that one. Just this morning, I asked a couple of our reps, I said, they know I play, you know, they know I'm spending a lot of time on AI and everything these days. And I said, I want you to use your imagination to create an AI tool that manages inventory without a human. And come back to me and see what you can come up with. And I said, well, you know, what are the limitations? I said, there really no limitations. Just use your imagination. You know, create some kind of Borgy science fiction kind of monster in your brain and see what it does. Because it's a, it is the constant nightmare in every device company.

Yep.

You don't know where your instrument sets are, and you've got implants parked in hospitals that aren't getting used, and it's just, you got money tied up all over the place. Yeah, it's a big problem.

Yeah. And the hospitals, the hospitals won't take any responsibility for it. The comp, the manufacturers don't, they don't want to take responsibility. Contracts are all written basically in the end of the day, the distributor gets screwed.

Yeah.

I'm so old, I remember when hospitals would buy the implants and instruments. I remember going to surgery in the '80s and you'd walk into the operating room, there was this giant— looked like a grocery store. There's this giant wall of glass shelves. And what size would you like today, sir? Okay. Oh, size 7 tibial component, right? And the nurse would pull it off the shelf and then she would ask to replenish it.

So I was a J&J dealer back in the '80s and I, you know, won the award for being like a top, the top dealer, so forth. And they said, what was your secret? I said, well, I hate to share this, but I figured out that if you do a total joint at a military hospital, they buy all the instruments and all the input. Smart. He did. And I, I learned it by accident. We did a case at Montgomery Air Force Base. And, uh, I got the PO for like, I don't know, $60,000 or $80,000. I'm going, what the hell is this? They bought everything.

Yeah, that's a good old days.

Yeah, then went to Fort Benning and—

remember the case? Do you remember the famous case in New York City where the— all the implants fell out of the ceiling in the operating room? That was— there was a Zimmer rep that was fake selling implants and he was stuffing them. He stuffed them above the tiles. In the hallway and the operating room. And so that there was no trace left. You know, it got used in a patient somewhere. And then one day the entire ceiling just crashed in.

I thought the story I heard on that was it was US Surgical.

Could be.

Yeah.

It was Leon Hirsch's group. And it was, I think, pretty sure it was a US Surgical rep.

They were stuffing the ceilings.

Derek Hughes.

Great talk. Yeah, good. Hey, Eric, you want to give us an update on your SI concept? What, um, or not your SI, your, uh, spinous process.

Yeah, not much an update. I've been, I've been gone. I had, uh, welcomed our first granddaughter.

So congrats.

Congrats.

So, um, but I do have, uh, I'm, uh, Talking with Andrea and Mark Copeland tomorrow to get some help.

Cool. Yeah, Andrea just asked me on Slack for your contact information, so I just gave it to her.

Yeah, I saw that. So I'll get going. I just need something else to do.

I think the biggest challenge, just from memory when you presented it, was just You know, there's market fit, you know, product market fit. Just, you know, let's just assume it works and you can make it cost effectively, you can get it through a 510. Does the market want it? That's the big question.

Yeah, the first part's easy for me. You know, I know it's better than what's out there and it's not hard to make. I can get 510 on it, you know, like easy. Um, but is it worth it? And I don't know, because there's nothing whiz-bang about that, that— what do you call it— a market segment, maybe. You know, it's, it's not, it's not that big a deal. Everyone has it, but I think it's just to say that they have it. I don't know if anyone's making money on them. The reason, reason it's, uh, out there still is because, you know, I developed it for Medtronic. I got it all done. They said, oh yeah, we're not going to sell that because it'll You know, it'll interfere with that pedicle screw business.

Cannibalize.

Yeah. I have, I have a drawer full of things that would interfere with pedicle screw business.

Companion Spine. That's where Companion Spine came from, right? This left island of misfit toys from Medtronic.

Yeah. A lot of it. You know, the big thing at Companion Spine was, was, uh, DEOM. So the spinous process. Uh, uh, the squishy spinous process spacer. Um, and then, then the Misfit Toys came in with that as well. Um, yeah, they, they also came along with it my, uh, percutaneous spinous process spacer, which I hope, hope they can do something with.

Interesting.

Yeah, but yeah, that's, that, that was motion preservation versus what I'm talking about is just, you know, adjunct diffusion. Okay, I'll let you know what I find out.

Yeah, yeah, market fit. I mean, Cope would be a really good person, you know, to start with. Um, you may, you may think about giving a really savvy sales and marketing person a little equity if they'll just do the sweat, sweat work and figure out if the market's there or not.

Sure. Yeah. Yeah. And that's, that's where I need the help of, because that's not my, my stuff.

Yeah, we all, we all have blind spots. All right, well, it's been an hour. It's been interesting. We're getting closer to— everybody's asking, why don't we just have a Mark Maloney hour? And we're getting closer to that.

This was it. I appreciate your patience on talking about this subject.

Yeah, and if anybody finds something interesting, just share it with everybody on Slack. You know, maybe you ought to create an AI channel on Slack, a depository for new tools and things. All right, thanks guys.

I'm gonna call my veterinarian and see if I got something for him.

That's right.

All right, thanks Kyle.

Thanks Tiger.

Thanks.

See you next week.