AI in Healthcare & Longevity Medicine: Why the Future Still Needs a Human Touch

The Cell to Systems podcast is for informational and educational purposes only and does not provide medical advice, diagnosis, or treatment. Listening does not create a doctor-patient relationship. Always consult a qualified health care provider regarding your medical conditions or before changing your health regimen. Do not disregard professional advice or delay seeking it because of something you heard on the podcast. Reliance on the information provided is at your own risk. Guest opinions are their own. Cell to Systems may utilize affiliate links, feature sponsored content, or discuss companies in which hosts or guests have financial or advisory interests. Relevant disclosures will be noted during the episode or below.

All right, welcome back to Cell to Systems. In this episode, we're covering AI in healthcare and longevity medicine. And boy oh boy, is this going to be an exciting one. Lots to talk about. We're going to talk about AI in drug discovery and peptide creation, or peptide creation as discovery. We're also going to talk about sort of how AI is being used in diagnostics, you know, also looking at some of the things that have happened with regards to some positives and negatives around that.

And then there's a new thing that I think is really, really interesting in York, Pennsylvania. The Wellspan Hospital has created a fully autonomous AI robotic kitchen to be able to provide meals 24/7 that launched recently. Why they're testing it in York, Pennsylvania, I'm not quite sure, but I think it's interesting that they made that available to people to provide healthier options. I think Suzanne, Leonard, Christy, Craig, you guys have all said at one point in time that in the hospital, you didn't always have the best choices for food. Right. For sure. Right.

Um, and then finally, we're going to talk about sort of some of the administrative things that can be done with AI in terms of automation. Certainly front desk stuff, things that we've worked on to create a better patient experience. There's nothing worse than being on hold and having to wait for staff to get stuff for you. There's a whole new wave of what we'll call customer service and capability through AI, Agentic AI, actual digital humans that can provide a better experience for people as they navigate through the health care system.

So, there's also been some bad stuff that's happened with regards to AI. People using it to create—I think there were some people that were hospitalized with regards to protocols that were created for peptides using AI that happened at a biohackers conference, I believe. I'm looking at Dr. Labarusha who's off to my right, which I think is kind of a scary proposition. I think once again leans back into the notion of what we always talk about: hey, talk to your provider, not what you hear on the internet or from some sort of influencer or whatever. And certainly, like, you know, getting your dosing schedules from AI, whatever large language model that is, might not be the best, best way to go.

And then there's also been some points in time where there was a product that was developed by Epic, and we all know that Epic is the hospital system of choice it seems these days. Craig, was Epic in your hospital? We do use Epic. Actually, we had been using a different system and I left the hospitalist role just before we moved over to Epic. Okay. So it seems like—Chrissy, was Epic in the hospital where you were? Yes. And Suzanne? It's been a really long time since I've been in a hospital. Okay. So, yeah. Okay. So, I think the reality is that there was some sort of issue with regards to AI monitoring blood lab results as well as vitals on a patient who went into sepsis because they were sort of using AI as the primary monitoring tool. So, that's probably not a great thing. So, we got a lot to cover.

And the first thing is I'd love to talk to Leonard and ask him about his thoughts around peptide creation or discovery. It seems like there are a number of things that are happening in that area. What can you, what can you elaborate on there? Well, it's, it's, it's very exciting. Um, it makes a lot of sense with AI just understanding the different—when it comes to peptides, understanding the different amino acid sequences and the ability for them to sequence that, you know, a million times faster than we as humans can test that. And the more data that we have, the more understanding we know the different cell signaling capacity with these combinations of amino acids, and their ability to predict what peptides might actually do with their advanced algorithms is exciting, just because more, more peptides are going to come to the market a lot faster. Um, I can't remember the name of the company, but I know Eli Lilly just, I think, either purchased or acquired a company that did exactly this when it came to AI and peptides. And so I think it's, it's, it's exciting. Um, I'm, I'm all for it. I think that we're going to just—it's just going to accelerate when we can put together, you know, new interesting molecules that can help people. But it is pretty early and I guess, I guess time will tell. We'll have to see, you know, what happens with it. But I'm, I'm, I'm optimistic on it. I think it's positive for now until, you know, we see what happens.

Yeah. Franc, what are your thoughts with regards to that? Yeah. So I think, I think, I think it's super exciting. There's so much advancement there with, you know, different AI models that are generating the sequence. What's happening out there on the manufacturer side is their ability to sequence those things. They're going to bind differently to different receptor sites because they're going to be able to actually see what the receptor sites look like. So there's a lot of ways to customize it. But the bad thing about it is, at the end of the day, every time you change certain things it does create, you know, side effects, unwanted outcomes. And that's kind of the part that can be a little bit scary, right, because yes, we're going to bring new drugs to the market, but at the same time, it's like we won't have a lot of information about what will be the long-term side effects of a lot of those drugs. It'll be fast on the market, but not enough evidence for safety data.

Well, you know, it's interesting. I wonder, I wonder just how quickly we're going to see a whole new slew of peptides. It seems like it's going to be, you know, exponential, but then again, maybe not. And how long will it take for that to go through the approval process? We're just trying to figure out what's getting voted on in July. Isn't that what's going on at this point in time? Yeah. Yeah. July is going to be seven peptides back on the docket with the PAC and the FDA to evaluate them again for safety, right, and efficacy. And then 2027 we're going to have another round, but we're in good shape, you know, and we're pretty excited about what's about to happen.

Is there a precedent that's already been set in this sense? You know, I mean, are there any drugs on the market, whether they're peptides or non-peptide pharmaceuticals, that have been generated or developed by AI? I'm not aware of any peptide per se, but I know for a fact that there's a few on the biologic side that have been, in a sense, designed utilizing AI. But I don't know of any new peptide in our space that went through that process. I know, I'm told that there's some companies that are using or leveraging AI quite a bit in oncology right now. None of them have been approved so far, but there's a lot of excitement on the market, on the stock market specifically, on what those companies are about to do. I think it's about to happen, you know. I think we're about to see a lot of those things hitting the market here very soon, especially now that there's, you know, a faster way to get things approved based on evidence. I don't know if you guys remember what we talked about last time with the way what the FDA was doing and accelerating, expediting new drug development, especially drugs that are leveraging AI, but I'm not think I'm aware of any peptides for sure. Great. Leonard, are you aware of any?

Um, I'm not sure of any in the pharmaceutical space. I know that there's companies that use an AI platform for a while now to develop OTC, nutraceutical peptides, and they've been doing that for a while. So, it's not necessarily anything new. But they've done it through the nutraceutical side; I'm not sure pharmaceutical yet. I'd imagine not yet just because, you know, the amount of time that it takes to get to market. I'm sure there's some in the pipeline, but supplements can get out to the public a lot, a lot faster than pharmaceuticals. I wonder if they're using AI to make those u-glow peptides. What do you think? No, they're using a kitchen sink. Kitchen sink. Okay, that's what I thought. Yeah, they just want to make sure that it's perfect, right? Just the right formulation, the right glow. Yeah, perfect.

You know, I think it's something that is going to be really cool is at some point in time. Dr. Fury, we had a conversation the other day and you were saying, and I've talked to Craig and I've talked to Christie about this as well, sort of the notion of AI inside of the EHR and being able to maybe not—well, I mean, at the end of the day, just kind of aggregate some information up for you that might be something where you say, "Hey, this might be something you want to look at." And I think that's exciting in clinical practice.

It's been fun. Right now, of course, everything is sort of a patchwork quilt. And so, we're bringing in little pieces. I think there's opportunities for growth in that area for sure. You know, we use in our practice every day, we use Freed AI, which is one of the AI scribes. It's a great product. There are a lot of glitches. One of the cool—you know, I had a patient this past week, a new patient came to see me with lots of complicated history, and because it was on a telehealth visit, it didn't record any of the meeting. And so I got to the end expecting I was going to have this beautiful transcript that I could use and send her this lovely, like, 18-item action item list, and there was nothing. So, I had to pull it all back together from my brain, which I had somewhat—I had turned off the, like, minuscule data brain and turned on the "what does the patient need" brain. And it was, it was a challenge for sure.

That doesn't happen super often. Most of the time it really works great. And one of the cool things is it captures all those little things like I'll say, "Hey, have you ever read that book No Bad Parts by Richard Schwartz? It's amazing. It would be really helpful in this particular situation." Or the patient will say, "While you're back there doing your propane injections, would you look at that mole that's over my right shoulder?" And I look and I go, "Oh, it looks like a seborrheic keratosis. If it gets any bigger, you know, pull it out." Well, 6 months later, she comes back and sees me. And I didn't write anything down about that mole on the visit because it wasn't what she came in for. It wasn't what I was on the backside of her shoulder for. And now I've missed that entire thing. Even though I addressed it with her and said, "Hey, come back if it gets bigger," it didn't get written down, so it didn't happen. Well, Freed doesn't miss that stuff. As long as it's actually recording, it doesn't miss that stuff. So, where I might have a two or three action item list if it's just me having a conversation with the patient, they don't miss anything. So, it pulls up all those little pieces of information that I went through and discussed with them and will put that together for me and it puts that into my medical record. So, it's a part of the chart. You know, I have to go back and adjust because sometimes it makes mistakes. Almost every time it makes mistakes, so I do have to review it. But I bet it's about 95% accurate. It is a really valuable piece of helper. My personal doctor uses the AI, uses Freed AI and a physical medical scribe. So, she has two things backing her up, which I honor.

You know, I have a Plaude device that I can use that records. And so, I could get my Plaude to do it instead of my medical scribe. Plaude has now a feature where it will do medical scribe for you. It'll put it in SOAP format. It doesn't communicate yet with my electronic medical record where Freed does. So, there's some pluses and minuses around that, you know, and there's all kinds of reasons to use all of those different devices.

Yeah. I mean, you know, the best part about it in my opinion is the fact that when you're having a cellular medicine or longevity medicine consult or followup, you know, you're hitting on a lot of topics and these are dense topics, and I can only imagine how difficult and how long those visits would really need to be if I was sitting there on my laptop. I'm not, you know, skilled at typing. I'm pretty fast, but I still got to look at what I'm typing. You know, if I was sitting there in front of my laptop trying to ask questions, take notes, ask another question, take some more notes—so the amount that it truly has freed my time and the ability to just really have a dialogue and engage with the patient in a very one-on-one fashion, you know, it's made this leaps and bounds ahead of where the patient experience was, you know, three years ago, right? So, it's, it's wonderful for the providers. It's also wonderful to just sit and be able to talk with somebody who, you know, is going to have a discussion with you as opposed to, you know, an interview.

I guess what I do notice is it keeps me from remembering details, though. If I'm actually writing it down, I remember the details. If I'm not writing it down, I'm paying so much attention to the clinical decision-making, the disease state, the treatment options, weighing the options, that I'm not paying attention to the whole picture of it all. So, it's—I mean, it's great because you don't miss that because it's now recorded in the history because that's what the scribe does. But I'm not holding it in my brain where I used to, which is freeing to some degree, but it's a, it's a plus-minus.

Yeah, we've been using A10 for probably the last 6 months or so. And it's been great for us because if anything, it helps magnify the importance of that clinical intuition, the pattern recognition, and then the human connection. But you know, I think one thing that we've all learned—you know, we learned so much through COVID, right?—but one thing that AI can never simulate is that human-to-human connection. Just like Dr. Fury, it's like we have because we're all using these AI platforms, that human-to-human connection, like sometimes I wonder if I have the same connection with patients that I used to have that I don't have anymore because, to your point, we're already like three steps down the road because it's already recording it for us. That's going to be phenomenal.

You know, I think longevity—like, we're in such a cool space. Our role is evolving every day. And we're becoming interpreters of very complex human physiology and, you know, we're—I just, I don't look at, I'm just not a prescriber. In my opinion, it's not AI versus provider, but the future is providers who understand how to use this and integrate it responsibly, and helping having technology enhance that intuition that we all have, that ability. We wouldn't be in this space if we weren't interested in it.

I think it's also interesting, like my learning style or my ability to write and such is I do best when I'm speaking. So I take Whisper Flow and I'll start talking to—I'll make a note or I'll do a whatever the thing is that I'm creating. So, for example, I'm going to create a Substack post and I'll just go into Whisper Flow and go and just talk. I'll walk around my backyard. I have this beautiful garden in my backyard and I start talking about whatever the subject is that I want to do a Substack on. And then I can take that and run it through this prompt that I created in Claude that has my voice and knows what I sound like and whatever. So it's this information that I have generated that's turned into this beautiful Substack that's in the, you know, proper best approach Substack format, whatever it is. So it's still me because I was the one who created the voice and I'm the one who created the information, but it's not me having to sit down with a pen and paper and write it out and then go back and edit it. I do go back and edit the Substack generated, but there's all kinds of other ways that I use it in that same kind of way. Because I speak better or because my ability to generate information is quicker when I speak, that's such a simple way to create all kinds of content for patients and for social media, too.

So, what is this? Whisper Flow. Whisper Flow, and it is an AI platform in and of itself that like sort of organizes your thoughts as you, as you speak them along? No, it's—Whisper Flow is, it attaches to all of your devices and all of your apps, and instead of using the microphone option, you use Whisper Flow and it will take what you say and edit what you say into like—so you don't have to put in commas and apostrophes. You don't have to dictate the comma or the period. It just does it for you. It's the most, it's the most accurate one. You know, I stayed away from all the voice record things because, you know, we're saying weird, weird words that don't, you know, that it doesn't know how to spell, but Whisper Flow, it's like perfect. I, you don't, you, you—I mean, you still read it, but you don't even have to read it because it just, it, it does it so well. And you can use it on your phone, text messages. I just downloaded it the other day on my, on my computer. Um, just, yeah, it's, it's great because I'm, I'm just like you, Suzanne. And I, I, I'm better at, at speaking, and it's, you know, you can get a lot of thoughts out in an imperfect way and you can ramble and you can repeat yourself, and you don't have to worry about it because you're just trying to get to that, that core point that you're trying to make. And I love the way that AI summarizes my confusing, backwards, not-in-order brain. It makes it sound coherent. So it's, it's fun.

And the, the same process that you went through, Suzanne, I actually spent a couple hours this morning doing it. The important part is that I guess you put a lot of time into your actual voice and, you know, how, you know how you want to write your Substacks. I think this is actually an important—not only for us as for efficiency, but this is something I've been talking to family members about and even, and even patients, in that AI is built to agree with you and it is, it's built to kind of make you feel better, and if you don't give it any context, it could have our patients go in a really wrong direction. You know, especially patients that maybe worry a lot or have a tendency to just overworry about things. It's almost like AI knows that and they kind of take advantage of it and they're constantly trying to reaffirm their things.

And so what, so one of the things I've been telling patients and some of the things that I've done is that you have to have kind of like an "about me" section or something that you put in the instructions or some type of document that you upload that tells AI about you and your specific situation so it has context. You know, what you do for a living, what your schedule's like, what you're responsible for, what your significant other does, what your goals in life are, because your answers are going to be so much better than just opening up chat and just asking it some questions. Because I've seen patients just go down the rabbit hole in really wrong directions when there's no context, where it's like, well, maybe if there was more context and they knew what your goals were or your, your situation within your family or what was important to you or what your weaknesses are. I think you should be able to mention those things because then the output's just so much better. It actually will remind you, "Hey, this is kind of one of your weaknesses. Maybe, maybe you should consider the other side."

I tell my AI to be like extremely mean to me. It's like, "Disagree with everything, you know, tell me I'm wrong, fight back, you know, kind of be annoying." And that's because that's the, that's the information that I want. I want it to constantly be, you know—it's, you can just go, you can stack little micro-inaccuracies on top of each other over a week's time or a month's time and it can turn into a real issue that you don't even see happening. And so it's not just how we use it for efficiency, but we have to start getting good at this because our patients are going to be out there using these, these large language models and getting their information from there.

I think it's pretty dangerous. And I do think that, you know, people go in there and ask all sorts of things. Now, you look at Gemini, we go across all the models. Gemini is certainly really big on disclaimers. "Hey, I can't give you medical advice. I'm not a doctor," this, that, the other thing. We've seen some things with GPT where that was maybe not as—you know, there's some bad information that was doled out there. I'm kind of curious as I listen to you guys. I mean, obviously there are a number of different applications. I have a Plaude as well. I love it. It's great. But I'm curious. Suzanne, you mentioned the, you know, the use of Claude. It wasn't that long ago that GPT was the flavor of choice. It was, it was the—since you're from Atlanta, it was the Coca-Cola, but I feel like Claude has won the Pepsi—you know, the Coca-Cola Pepsi challenge and is the flavor of all people at this point in time. Any of you guys want to comment on that?

Yeah, I was—I, I used a lot of ChatGPT, and in the past maybe two or three months, just Claude one day from one day to the next just got so much better than everything else and, um, just it was just kind of like happened overnight. And, you know, sometimes a lot of your information is on the previous LLM that you're using. And so for about a month, I was asking ChatGPT, "Hey, tell me everything you know about my business or my thing on this subject." And then I would copy it and put it into Claude because just the responses were so much better and the integrations were so much better. And the education around how to use it more effectively was, was so much better. And it's, it's just a crazy time because by the time you figure one thing out, there's an update that just kind of eliminates a business. I'd be very scared to be in the tech world right now building something around AI because tomorrow it might be an update on Claude that does it with one click of a button. And so it's fun for us, but it's going to, it's going to be kind of scary for a lot of people at the same time.

Yeah, I was just—I would just say as a person who's in that business and develops AI technology, I could just say that there's also a ton of stuff that's out there that's vibe-coded. It's not good code. And so at the end of the day, there's a, there's a lot of activity. We'll see the cream rise to the top. And so it's those people who know how to develop technology, been doing it for a long time, are going to be the ones that typically have the process methodologies in place for QA/QC. You know, if you think about it, Amazon got rid of a bunch of junior developers, mid-level developers, to make way for an AI infrastructure. The AI infrastructure essentially pushed a line of code live. So, normally it would go into a staging environment. It went past the staging environment, pushed it live, and took Amazon's website and app down for six hours, which cost them $631 million. So, some change for them. Yeah, I mean, I think it probably didn't put too much of a, a dent in Bezos's pocketbook, but nonetheless, I think that there's, there's a lot of danger in the reliance on just thinking that all of these applications really are doing everything the right way. And there, you know, some of the implications on the healthcare side could be significant. So again, I think it's, you know, the points that you're making, that you guys are making around really paying attention to using those products are super important.

A different animal, right? I mean, we, you know, one small mistake is not—is, is going to be very costly, which—someone said this and I thought it was really interesting, and see what you guys think, that in the next year a doctor will get sued for listening to AI and making a mistake, but at the same time, a doctor will also get sued for not listening to AI and making a mistake, and that might all happen at the same time. I thought that was a pretty interesting thought to think about is that we're in this transitional period where are we, you know, are we liable for not listening to the recommendation? Do we have any protection if that recommendation came through? And like you said, Jock, in healthcare, this is a completely different animal than some of the jobs that might get lost to people that were doing design work that now we can do in two minutes that used to cost, you know, hundreds of dollars an hour, which I'm enjoying right now, not spending anymore.

Well, and also the—you know, it's been probably my go-to for years and years. I even lectured on it. I'm sure when, when Leonard, you were talking before about being in a lecture, I'm sure I lectured on using Google for my research for all that stuff. And now what I use is Perplexity and Open Evidence and Google, because they give sort of different perspectives. Each one gives you a little bit different information and you can be a little bit more flexible, but then there are associations that your brain can make that those can't make somehow. And so you still have to do your own research. You still have to go look for the thing because you're like, "I know there's a thing that makes this work." And so then you have to go back and do your own, you know, when you're, when you're putting together presentations for different things.

There's so many great tools. I know, Leonard, you probably do that as, as often as me. And, you know, Gamma and NotebookLM and the beautiful, beautiful slides that NotebookLM can put together, and then going back to Google Slides and kind of the beautify—the new beautify option on there. It's so, it's so beautiful, the things that you can create now.

It is. And but you know, there's, there's things that I've learned from, from you, Suzanne, that, that are just like these small little points about—I don't know, like a butyrate or some, some of these, like, key little clinical pearls that—I'm in Perplexity and I'm in all these places and I'm, I'm, I'm accumulating all these studies and I'm saying, "Hey, there's a connection here," and they're like, "No, there isn't." I was like, "No, there is." And I'll have to go back to some of our old notes, our old slides, and say, "Look, this is the title of the slide. This is where they, they clearly saw that there was this mechanism within butyrate." And they're like, "Oh, yeah. You know what? You're kind of right." And, and, and that happens all day, every day. And so it's exciting, but it's also a little nerve-wracking because when we lean on some of these models to put together like a nice slide, I, I'm, I'm always like so nervous about that one sentence that they kind of hallucinated on or, you know. And so it's like I spend more time using AI because now I'm putting in my own two cents using AI to summarize, but then checking back on, on, on the AI's work because it's, it's not ready yet. Like, I'm sorry, it can't replace some of the brains of the people that we have on our science, as a scientific team. But because, because of learning from them, I can argue with AI and say, "Nope, you got to look a little bit deeper. There is a study that says this exactly." And usually they, you know, they, they say you're right.

But, um, it's—we're actually for, for Nucleus, our education platform, we're—I just started—I'm not starting from scratch, but we're redoing everything just because there's some really great benefits to AI when it comes to education, but there's going to be kind of like an ego thing that happens at the same time because we're going to have to take all the things that, you know, our, our assumptions, all the things that we thought we were right on, and we're going to have to run it through AI as well to say, "Hey, push back on me here. Where am I making something, you know, something sound evidence-based where it's really more mechanistic plausibility, you know? And am I, am I crossing the line by, by talking about a pre-clinical model, something that happened not in humans?" And the way, am I describing it, does it make it sound like it's really evidence-based? Because we want to re-go through everything we've ever said, all of our documentation, all of our slides, because we just have access to so much more information now. And while that's fun and I'm, I'm looking forward to doing it all, all kind of revamping the whole website, at the same time, it does make me a little nervous because on both sides of the spectrum—but it is, it is fun and I feel like we have more information at our fingertips and more of an ability to help medical practitioners, um, with gaining the confidence that they need to, to help their patients.

One of the things that I—I don't know if it's happening in your guys's area, but these big national wellness firms are coming into these box gyms and are offering them these, these wellness packages. And I actually had a patient—I don't know if it was this week or last week—but that just—it, it was clearly it was just a copy and paste optimization protocol that they were given because, like, you, these firms, and I'm sure they're just AI-generating people. They're just chasing these protocols instead of really understanding the physiology of, of what's going on. And I, you know, to me, that's what makes our space—you know, we can't be replaced by AI. Can we be enhanced and it help us with our processes and understanding and our workload? 100%. But, you know, it's, it's asking why the body is compensating in the first place. I mean, I think that's what makes us so good at our job instead of just literally going through this code, you know? And it's just, it, it kind of frustrated me at first because I was like, "Really?" But, you know, they're offering these optimization, you know, peptides that who knows where they're getting them from. But the, the, the patient's not going to walk away with improved health span. I mean, we all, we all know that. And, you know, I mean, the future of medicine is, is not less human. It's just more precise.

Yeah, it's so great, Christie. One thing I want to say about that is you think about it, it's kind of like going back to things we've talked about previously. It's the Hims & Hers, like this one-size-fits-all kind of like—you put it in, who knows if it's AI or not, right? Who knows if that's just some sort of, you know, machine learning algorithm. But at the end of the day, you all, when you see patients, are basically seeing the whole patient, and it's not some sort of out-of-the-box, "here's a protocol," like you said. So I think the notion of providers being replaced by AI in any way, shape, or form seems almost ridiculous. I mean, it's one thing to make a meal, right? You have all the things in a row and like—there's a certain ounce of this, an ounce of that, an ounce of this that goes into the meal. But for what you guys are doing, it's just a totally different, uh, ball game. So, um, I think that's one of the things that's super interesting.

My daughter is obviously—I think I've shared with you—is all studying for, to be, ultimately either a, you know, psychologist or perhaps a psychiatrist and, uh, wants to provide therapy to people. I said, "I think that's one of the greatest decisions you could possibly make because as we change in a world of AI, universal basic income is probably a reality and there are going to be a lot of people that are going to be feeling kind of weird about not working, sitting around, jobless." And that may be, you know, kind of far out thought process. But you know, to Craig, you alluded to having, you know—so before when we were getting ready, some people were having actual relationships with AI. I don't see how you could actually have a relationship with an actual, you know, AI psychiatrist and how that really makes sense. It just doesn't seem like it would work. Does anybody think that, that I'm crazy?

At least, you know, at this, the point where the technology is now, it certainly doesn't make sense. You know, we've probably several of us have seen the movies where, you know, Joaquin Phoenix has found an AI girlfriend. And, you know, I think down the road, you know, who's to say the sky's is the limit. You know, there could potentially be some, you know, room for developing a relationship with a psychiatrist, psychologist, that sort of thing. But, you know, currently where it is now, it, it seems very, I don't know, diametrically sort of opposed. You know, human tech.

People are marrying, I think, their AI chatbots right now. So, I think a therapist is probably in the cards. Yeah. I mean, people are very lonely, you know, and they just want to be cared about. And when you're chatting with ChatGPT and it starts to get to know you a little bit, you know, I don't, I don't blame people. I mean, I can see people connecting in a way. I've heard people talk about their chatbots—and educated people with friends—and, you know, so I, I think there's something there and I think it'll get better. It just doesn't feel very human or natural for us to, to think that that's okay. But I think that, I think we're honest that positive for people and I think it'll only get better.

I have a question, uh, for you guys. Um, just going back to something that we were talking about earlier but also, you know, talking about this relationship that one develops with, with their LLM or, you know, how well it gets to know you, your practice. You know, I used to love Claude and, and Anthropic, and I was a big Claude user for a little while. And then I sort of navigated away to ChatGPT. I was following these, you know, LMS arenas where you can day-to-day, hour-by-hour sort of track the performance and, uh, across a number of different metrics between these, um, learning language models. And, uh, so you know, I've, I've stuck with ChatGPT, and now it knows me so well. Like, it knows my practice so well. I'm sort of reticent to move to any other, you know, company or, or system because I feel like it's going to take so much time to build that back up. Is that—is that a fallacy? I mean, is that—

No. I, I think the reality is that the bottom line is there's, there's a way to do it faster than the way that Leonard did it, which is you basically—I'll show you when I see you in June. Yeah, you can basically download your entire, uh, thing and move it over. But the one thing I will want to caution everyone on is remember, um, you know, this is healthcare and where your data goes matters. So that's why I think what's happened with, um, you know, Nvidia using the Llama models to create all the Neotron stuff that they've done and sort of these sort of siloed—they're not open, you know. So basically what, um, Nvidia did—you know, Google was really hot on owning their own supply chain for AI, so they have it all. They have the software, they have the TPUs (tensor processing units), they have the actual data centers, etc. So long story short, they could control everything that they needed, and that's why Gemini has just taken off and done so well. Everyone is dependent on the GPUs that come from Nvidia, and so since there's a shortage of those, um, you know, there's a lot going on.

But one of the things Nvidia wanted to do—Jensen Huang wanted to make sure that they stayed in the game, so they didn't hitch themselves to any one wagon. So what they did was they basically created—they used the Llama models, which are open source from Meta. So, I don't know if you guys have used the Meta AI at all. It's pretty good. Um, so they use those open-source models, created their own models within that—within their—so their own sort of large language models, um, that are based in their own environments, which is pretty cool. And those are open source. And the idea is that you can then move all of your data into a spot where it belongs just to you. So there's a lot of stuff that's now, you know, sort of like micro environments, a little bit larger environments. So we have like cloud code or, you know, open quad, these environments that are running on Mac Minis. We also have those in the cloud. The list goes on. But, um, I think one of the things to think about for healthcare is where will your data live, um, and how do you protect it? So, how many people are really worried about that? Because I, I hear that statement all the time. Um, you know, "where is your data, who owns it?" And I hear that a lot, but, um, do people really care about that? Because it seems like they're just giving away their health data and putting information anywhere without really thinking about it.

Well, I think it's different from being a consumer and being a provider is my point. Yeah. Well, do I know the providers think about that. Do you think the consumers are, are thinking about that? I don't think—I, I think at the end of the day consumers are like—if you think, let's relate it back to social media. When we put stuff in social media, they own it; the minute you put it in there, it's their property. So I, I, I don't think the consumers are as concerned about it as—they're probably putting in tons of information. And you know, um, one of the things that Anthropic was really against was like this large, you know, mass surveillance capability of, of these, um, of these models. And so, you know, I think at the end of the day, um, where you put your data is up to you. Uh, people may or may not care about it. I think it's something to think about, though.

I know for, for us, um, Leonard, just like when we first introduced A10, I have a consent, so it's part of when they come in, they have to sign the consent so that they—we can use it in the rooms with them. And, and it was just—it was actually a, just a staff education that, um, I didn't educate my front staff well enough to explain to them that it's—this isn't just an AI program. But at first, that's how my staff was explaining it when they were going through their consents to sign. But, you know, now that they're explaining, "No, it's a way for us to collect your information better so that we have can have a clear, more concise, um, picture for you at the end of your appointment." So, that was just my opinion, as far as—I do think some consumers, like, because at first, all of a sudden we were getting—they, you know, denied the consent for us to use it in the room. But it was just merely a staff education on, on my part that I didn't—I didn't see coming until I started seeing it.

Yeah. I mean, I think it's super interesting to think about like our product Grace, which is Genuine Responsive AI for Care and Engagement. We spent years developing this, you know, so that it's a digital human. It can literally have a conversation with a patient, um, schedule an appointment. It is not just a chatbot. It can actually control your system. So Craig, you had asked me about some automation things—we'll, we'll talk about in greater detail—but how do you automate things within the practice? And that's one of the things that's just so amazing right now: the notion of having a front desk that is, you know, available 24/7 that can get, you know, a patient scheduled, find the right, uh, information for them, but also has off-ramps, right? It's within guardrails. So, it's not in this place where it's sort of like it's going to answer whatever the patient asks. And, you know, we've had—we've tried everything we can do to break it, to find, you know, what's, what's, you know, what's wrong with it, what will it do, where, where will it, you know, mess up.

And the good news is if you really spend the time to do it the right way, you can create a product that actually is pretty bulletproof and can create a better patient experience. And I think that's what, you know, patients want. There's nothing worse—I mean, we've all been on hold, right? There's nothing worse than being on hold or, you know, trying to, trying to book something and it's not quite right or whatever, or even at the notion of, "Hey, there's an abnormal lab that came back or something," and then you need have it call out to the patient, uh, to, to have the, uh, the patient come back and speak with you. So there's, there's a whole bunch of new stuff that's going to happen from an administrative standpoint on the AI side, not just the clinical side. I think that's super exciting for, for, you know, practices that are trying to provide the highest level of care and be as efficient as they possibly can be.

Yeah. The, the administrative side, um, is—I, I, I think I kind of made a post about this and I think someone took it the wrong way, was that, um, you know, I was saying if I was running a practice right now, I'd really focus on what you just described, Jock, which is, is the ability to improve efficiency in administrative processes, um, so that it would free up the physician's time to create more content. And, um, and the reason I said that is because you only get so much time with a patient. You might get—you know, some people get 15 minutes. Some people get like an hour, an hour and a half, but that's it. And I, I, I do think in the future of where we're going is there's, there's trust with the physician and, and the patient. And, um, you know, giving the, the, the, the physician time to do a podcast or time to do videos—not because they want to have a, a massive podcast or be famous, but it's so they can communicate with their patients and have a library of, of, uh, of topics like the ones we've been talking about, but one on sleep, one on cardiometabolic health, one on nutrition so that they can focus there. And it's almost like they can give their patients 20, 30, 40 hours of themselves.

And, and I think that that's going to be the future of, and what you're talking about, Jock. But to, for that to happen, for them to free up their time, they're going to have to improve efficiency with these administrative tasks that you're talking about that frees up their time to, I believe, have even more of a relationship with their patients. And then someone got mad at me and said, "Leonard, you're completely off. Like, get off the internet and go, go see and go spend time with your patients." Like, "No, no, no. The point of it is so that I could get more information to the patient, not to like ignore them and go create content."

But yeah, we're in the middle of an, of an office visit with a patient and I've got my Plaude on. I do this now all the time, or in my staff meetings or my provider meetings, and I'll just hit my Plaude when I, I feel like I'm starting to teach, you know, I get that like teacher comes up. Yeah. And I go click and it records everything I say because all that information needs to be in my, um, brand guardian. It needs to be, you know, used in my Substacks. It needs to be used because that's all valuable data that's all coming from my brain. So, if I can take all that and put it into a transcript, that's where it's valuable. And then you can chop it up into little pieces and your patients can—instead of me having to say the same thing or draw the same drawing over and over and over again, I can, um, actually be giving that to them and record it one time and then say, "Hey, watch this little video. Bye. Be right back."

That's awesome. Now, what is that? Plaude. Plaude. P-L-A-U-D. P-L-A-U-D. Like applaud. It's a little like a—and that's the, that's the one thing that no LLM, no Claude can update. Nobody's going to update, nobody's going to update, uh, Dr. Fury, you know, last 15 years of her experience, you know. And that's, I think that's what physicians should be building, is, is this library of their unique process and how they do everything, because nobody's going to be able to kind of compete with that.

I'm so glad that, you know, I don't, I don't, I don't need to be right, but I like it when I am. So seven years ago when we started building Quantum, you know, to just do this exact same everything that you guys are talking about. It's just so cool to hear you guys say that the, the notion of being able to give the, the time back to the provider as well as, you know, provide a better patient experience. Back to the Plaude, it's just this right here. Craig, you put it on the back of your phone. Although Suzanne, did your—was your original like the wrist one? That's what I had originally. I have a little pin, a little like note pin. It's like a—I don't know, it's the size of a small egg.

Yeah. And there's the—there now there are these new, uh, like wearables that basically record everything visually as well as, uh, the audio. It's just incredible. The technology is definitely taking off. Um, and you know, just again I think is trying to figure out who's the right company to work with and who's going to be around because the other side of this too is I had a really interesting conversation with somebody—obviously this is kind of the area where it's all kind of happening—and had a really interesting conversation with a friend last night and, you know, there are, uh, trillions of dollars that are being pumped into, into the AI development market. It's kind of like when we were here in the .com days, there was this huge rush. And at the end of the day, you know, somebody wants their money back. So, we're going to, we're going to see a point in time where stuff that's free, you know, you're really—if it's free, you're the product. Uh, so the long story short is that, you know, we're going to get to a point in time where, like, you know, Suzanne, you mentioned Perplexity. Um, we have a Perplexity free model. You got the pro model. You got the premier model. You have all these different models that, you know, cost different. It's like Grok, right? Uh, Super Grok, $300 a month. Well, at some point in time, these things, these subscriptions, are going to be pretty expensive. So, I think it's going to be interesting to see where it all goes and who wins the war. Uh, with regards to that, I think who wins the war is the one that's going to continue getting as much data as possible for the least amount of money.

I, I mean for me, just like, um, it's, it's when I think about AI and longevity, I, I think about, I think about the craziest thing ever, and then I just try to think about my brain go to next level of what crazy can look like times 10, because it—because anything is possible. And, and I, um, what's crazy for me is what you just said there, uh, earlier, Jock, was around data, uh, your data. And, and I think that's for me the big play there because, um, as customers and providers, patients are out there putting information because at first, you know, when you go to an AI agent and you, you put your information, you're scared, right? You're like, "What are they going to do with my information?" But and then, and then you really like what's happening, what you're getting, you know, the response that you're getting, the information you're getting, right, uh, the direction that it's helping you take, and then you put more information into it because you're like, "Oh my Gosh, I love this. This is good stuff. It's helping my practice, helping me. It's helping my doctors. Here it is."

And then you're going to start putting things very personal. Then you're going to start putting things that are extremely, extremely personal. And then you're going to start putting things that are intellectual property, such as contracts and things like that. And what's happening is you put all your information in there, your health, your business, your life, everything in there. And then you can't get out of it because it's already in it. Everything is in it and then you're like, "Oh my gosh, I just gave this thing everything I have, right?" And then, and then you realize, "Holy shoot, I now—I'm really dependent on this thing." Right? What happened if they take it away from me? Claude went down for a couple hours the other day. I thought I was going to lose my mind.

You're going to lose it. It was like—and, and as I was flying back from, uh, Mexico, um, I just listened to this thing on Delta that just—a very simple 10-minute crazy guy talking, um, on the Delta thing. And, and he said something very interesting. He said, "Well, the AI agent, which in this case was Claude, knows your name because somebody uploaded your name into it trying to figure out information about you. It knows your data, your health information. It knows everything. Someone mentioned your name into the system." So the system knows who you are as a person, as you're healthy. Are you healthy? Are you sick? What's wrong with you? What did that doctor ask about you? What did that business partner say about you? What did this, this potential new business is asking about you? It knows everything about you and, and where it's going. It's like what happened if that information about you becomes the way people deal with you? Is that fair?

Think about it. Think about me going to say, "Chrissy, you know, I don't know who Christy is." And I'm like, "Okay, let me go to Claude and say, I want to do this thing with Chrissy. I want to, I want her to be my new doctor. Tell me everything about her. Do you think she's going to be fit for me as a doctor-patient relationship?" And, and, and the AI bot knows everything about you and knows everything about Chrissy. And then he's got a line and say, "Oh, okay. And I think she'll be the perfect match for you." Or, "I think, no, you guys are not going to align because somebody said something about Christy and put that in the system somewhere about her, right, and you might misalign with what you want to do, right?" And, and, and, and that's what I'm scared of, um, but it's also exciting because it can also tell you about where we're going because you might be able to connect people that otherwise will never be able to connect in a sense, right? I don't know. Personally, I think it's Great. Um, the data side is what scares me, right? Yeah. What do you guys think about that part?

As a, as a consumer, on the consumer side of things, you know, myself and several of my friends and colleagues had done the 23andMe back in the day, and then they had that whole, you know, fallout or they got bought out by somebody and, um, you, you had the option to basically dismantle your whole profile and get rid of your data and all of your biologic material and so forth. And I mean, you know, pretty much all of us did. So I think that there are definitely a large number of people out there on the, you know, consumer side of things in healthcare that are, that are interested in protecting their data. They're curious about where their information is hidden, um, and who has access to it. So something that we all have to keep in the fore of our minds for sure.

The other place where it frees up, especially for us small business owners, right?—like I have a team of 10. I'm sure you all have very similar size practices. You know, we're not, um, uh, and so for me doing things like creating an SOP, creating an email sequence for patients who are new patients to the practice to introduce them to the practice as they come to the practice, and then when they leave the practice having an email that goes out to them that says, "Hey, you're amazing." But all that can be created, right? So you do all this work to put together all the things that you're, that you're talking about, Franc. And you put all that around the business. That's what I have in Claude right now is all the business stuff. And so you put all this time in building this giant prompt that is who Vine Medical is. And now you're, um, now you can create all this thing, all these things from it.

I can do, um, weekly staff meetings. I can do that—just take this—they take this long to make and it creates action items for me. I then record the staff meeting on my Plaud or on my Otter, take that transcript, put it into the thing, and it makes a new staff meeting thing. This used to take me hours of not medical, not educational time. It was—this is like a game-changer for me. And I know this is like brand new for everybody doing this, but for these tiny little practices that most of us have in functional medicine or in cellular medicine, longevity medicine, it is so amazing to be able to, to get rid of all that time. So I do what I was saying earlier, walk around my backyard with my, whichever, Whisper Flow or whatever it is, and I record, I talk about my employees. Okay, so Melissa has been with me for 11 years and she's really got this thing and she does this. Oh, she does that thing that annoys me. That same thing I was talking about earlier about content creation, it's the same thing with your employees. So, I don't have to sit down and type into a form, and it comes out very personal, uh, and it takes that long. It's really amazing once you build the thing. It, it's really amazing to have that, um, administrative help that I haven't had before. Yeah, that, that is phenomenal.

You think about that. I mean, there, like in the agency world, um, the, the big rush has been, you know, to be a frontier firm, and you see that across the, across the board people, you know, trying to create as much efficiency as you possibly can. You're just better able to serve the client. Like Leonard said before, hey, you know, some of the, the design stuff, Suzanne, some of the things you talked about, beautify inside of Google Slides, etc., you know, things that you can do. Uh, at the same time, I have to tell you, over the last five years, it's been extremely frustrating at times, like, you know, where we're trying to actually get it to do something and it just won't—like the hands come out all, you know, weird or whatever, and it just does stuff that doesn't make sense. And so, there's a lot of stuff out there that can actually save you time, a lot of stuff that can actually create, uh, the need to go back and, and look at stuff. Like Leonard was saying, I, I absolutely agree with that. Half the time whenever I do anything, I've got to go back and I've got to read through it.

And here's the most frustrating thing—I don't, I don't, I want to make sure I spare you guys all of, all this. If you ever use Grok and you have a conversation with Grok where you want to actually, you know, do that sort of thing like you're having a Whisper Flow, but you're talking to Grok, uh, and, and then you want to save the session. I was—I had a 4-hour conversation with Grok just going. I mean, we're just going and it's, we're just—it's so great. I go back the next day to talk to this, says, you know, start the project back up again and all of a sudden it's like, "Oh, sure." Yeah. And it, and it was lying to me. It had not recorded, couldn't, couldn't keep it, uh, couldn't log it. And so, I mean, it was like—I think I'm still upset about that.

So many things to talk about, guys. It's been an incredible episode. Thanks everyone for joining us on this episode of Cell to Systems. Please remember to like, share, and subscribe, uh, especially on episodes where you feel like you got some information that was really valuable. Sharing that with other people or leaving a comment, liking it for the algorithm, really helps us here at the channel in order to get the information out to more people, which I think now more than ever is super important, uh, because there's a lot of misinformation out there. We'll see you on the next episode. Thanks a million.