1241: Garner Health: Revolutionize Healthcare and Beyond Using AI, Prompt Engineering, and Go-to-Market Strategies with Marketing Strategist Rebecca Shaddix

Robert Plank 0:01 Hey, we're back with Rebecca shaddix, who is a seasoned marketing executive. She's a Go to Market Strategist with a proven track record of driving significant revenue growth across various industries, and lately, she is very interested and very knowledgeable and helpful about the societal impacts of technology, especially AI's role in solving global issues and fostering diversity in tech. You want to look her up on LinkedIn as we're speaking. Her last name is S, H, A, D, D, I, X, it's Rebecca Shattuck, the one and only So Rebecca, glad to be speaking with you.

Rebecca Shaddix 0:38 Great to be here.

Robert Plank 0:40 So we hyped you up a little bit, but in your own words, what are you currently, just like, really fired up excited about

Rebecca Shaddix 0:46 I am super excited about this, the potential and the pace of technology to solve really challenging problems. It's what got me into tech initially. But I think the pace now of being really critical about asking, what are the problems that tech can solve at scale is just never been faster in my lifetime. So I think it's great to see different types of challenges that different people are solving and combining a lot of those innovations.

Robert Plank 1:12 Ooh, I like that, combining idea and so we're approaching the singularity, and it's easy to get afraid of it, right? And think like, well, what if AI's outsmart us and like, we're dumb compared to the AIS. But then you think about all the amazing things that it's all this, like, cutting edge stuff, isn't it? They say like, they can detect, like cancer sooner, or they can like, better route, like airplane traffic, all these problems that they used to say, like, like, you know, logic wise, were really difficult to solve. But I'm curious about, like, like, what about these practical uses excites you? You say that, like, something drew you into AI in general, and there's this, like, combining idea. So, like, I mean, what can ai do that is really newsworthy and noteworthy today?

Rebecca Shaddix 1:57 You know, I think we ultimately can't predict right now, even in the next 20 years, let alone our lifetimes, what it can solve. But what's exciting is just this promise that there are these really complex challenges that we may not have had an ability to solve before combining now pretty quickly. So in short, I'd say there's very little that we can conceive of that I don't think that eventually tech and AI will be able to solve for right now, just seeing that pace of innovation, how we can have an impact at scale, decide on resourcing things differently at scale, looking at large quantities of data that humans have a hard time understanding or processing, to find patterns, to find solutions, even diagnoses like you've mentioned, This is all becoming very possible.

Robert Plank 2:42 And as far as where you fit in, like, like, what are you doing with AI or like, you know what specifically andlike, what industry, what solution are you kind of like, getting involved in?

Rebecca Shaddix 2:52 I work in healthcare marketing right now. I've worked in education. I've worked in HR software. And so the power of technology to solve these really big challenges in education, in HR, in healthcare, we're seeing right now, the pattern recognition is really strong. So the company I work for Garner health using data on real patient outcomes to figure out which doctors have the best impact on their outcomes, is hard for a single person to do. We can't process URI 60 billion medical claims and figure out what each individual doctor did for their patients, but AI and large language models can and so that's where we are now.

Robert Plank 3:32 And I'm always curious about the like, the challenges, right? The difficulties, the struggles. And so I'm always amazed at how, ever since the the large language models became mainstream, how you can kind of, like, type something and kind of mess it up and have typos and say it on your phone and, like, somehow it still knows, right? And so it's, it's really crazy fascinating idea that, like, different kind of types of data can somehow fit together. Like, my understanding is a lot of the like, the the healthcare problems and like, with like VA and things like that were, there was problems with, like record keeping, right? There was like, different file formats that were incompatible. And they were like, this is this huge problem. Like, you know, millions of records, all these things, and we can't fit them together. And then with, with the generative AI and the llms, like, kind of it can somehow figure it out. And so as far as, like, you working with that, like, I mean, what's what's changed the last few years as the technologies developed, as far as what you're doing in healthcare,

Rebecca Shaddix 4:34 it's a great question. I think the really actionable outputs from all of these inputs has really been what's changed quickly in the last few years, we can use better analytics now to evaluate the impact of single physicians, etc, on patient outcomes in a way that just isn't possible without an ability to consolidate and make sense of that data. So this amplification of what humans can do, what data scientists can. Do with the best of technology is just rapidly increasing now. So the conclusions that we can draw confidently, transparency into the inputs into those conclusions, all of these, are just getting stronger. And this is true across sectors. This is true making conclusions across healthcare, for sure, but also education and even HR understanding the impact of specific managers on employee engagement. All of this just is becoming increasingly possible to do, and really amplifying the abilities of people, not replacing them. So the people who understand how to leverage the technology and then use their unique insights in ways that AI isn't good at AI isn't good at a lot of the creative, complex problem solving that we are it spits out outputs through the patterns and the inputs it's ingested. This combination is just making people able to make such more confident conclusions and better decisions.

Robert Plank 5:51 And I think that's a really important insight there, that there's AI doesn't do everything, and there's really a need for like, these new skills of understanding the AI and like, having the right mindset, and like, the right strategies for figuring all that out. And so, like, you've, you've been through all these different kind of industries, right? You've been, you say, you've been education, HR and medicine, and as far as, like, your own personal journey, and like, what you've had to figure out, like, what's that been like? Like, what kind of like, you know, uphill battles have you had to fight on your way up?

Rebecca Shaddix 6:24 You know, I think the questions with AI are similar to the questions I was asking a decade ago about data. Really can't make decisions for us. It'll take the inputs we give it to spit out outputs. But if we don't ask the right questions of the data, even 10 years ago or at the right prompt engineering now, we can't be confident in what it's spitting out, because it's really answering a question that we've put in. So the uphill battle, I think, for both, is that there's a relatively low floor to get lots of dashboards and marketing and that lots of data. There's a relatively low floor to use AI in some capacity, but a pretty high ceiling to actually understand if you're asking questions that are yielding actionable outputs. So I'd say that's been the biggest learning is, how do you engineer prompts effectively make sure you're feeding in the right data, the right inputs, being mindful about the conclusions you're drawing from it, and then definitely double checking and amplifying with qualitative insights and just your own critical thinking from the outputs spitting out. So the prompt engineering, I'd say, with AI specifically, has been the biggest learning curve and the most rewarding figuring it out to get better insights. But the same thing was true of analyzing dashboards 10 years ago. You can't just stick up a dashboard and say, Well, what does the data say? And let that be a decision. The understanding of what goes into it has always been important,

Robert Plank 7:45 wonderful. And, you know, the prompt engineering, I think that that could be really helpful for us to talk about, because, like, I find prompt engineering to be it's kind of a lot of the upsides of programming, but without so much of the frustration, right, like you have one semicolon, one curly bracket out of place, and it could take hours, days to figure that out. But with prompt engineering, like, a lot of the times, the output is not quite what you want, but it's close to it. And it's a lot of fun to think about of like asking the right questions, or kind of like guiding the data or putting it within certain constraints and saying, like, you know, we'll do this processing, and then here's the kind of the language part, and then processing again. And so if someone is looking to understand prompt engineering, and especially the way you do it, like, what's your approach to it? Are you like a framework kind of person, or do you have a certain, like strategy mindset towards getting it right.

Rebecca Shaddix 8:41 I do. I say it varies a little bit by what we're doing, but ultimately, the framework is giving a very intentional role or scope. You are this journalist, this copywriter, trying to answer this question with this data, using nothing else other than what's been provided or doing research from these sources, answer these very specific questions or yield a very specific outcome. If it's just make this better, write a strategy, answer a big, bold question. There are certain questions that can be open ended, but be mindful that that's probably a more controlled input that you wouldn't want to then extrapolate beyond to make broader conclusions. So being really mindful and intentional with the inputs, the requests and the wording, because it's just going to spit out like a calculator, exactly what you put in. And so if you're not careful about the scope or the role of the inputs that are effective, that's where people can go off the rails and not realize that they're asking for something that they don't mean to be yelling,

Robert Plank 9:44 yeah, garbage in, garbage out, right? That was said a lot of right around when chat GBT became mainstream and people were trying to get their heads around, and it seems like you hear a lot less of that and and what? So what I'm getting to in your answer is it can sometimes be interesting to just have it make some. Up and, like, ask an open ended question. Ask something Wow, just as as part of the process of figuring out your kind of more detailed prompt, or your more specific output, like, you can kind of have these steps towards getting it figured out. And so, you know, we've been saying that, like, humans can be sort of the weak link in this sort of thing, right, like, they misuse the tool, or they're not involved in the right part, or they don't, yeah, and, and so, and how do we solve that problem? Like, it seems like AI has been around for like, a while, but only recently. Like, there's all these tools are just, like, talked about a lot now. And so how do we get the adoption to be better, and how do we get people to, like use these tools more properly,

Rebecca Shaddix 10:42 ultimately, by seeing them as tools, I think, and not a replacement or an abdication of responsibility, it can't draw strategic conclusions without the right inputs. It can find patterns, and it could yield the next prediction. But if you're expecting to just abdicate messaging strategy, any kind of marketing strategy, any kind of big, creative, strategic, complex decision, that's where people will bump into trouble. So, small questions, brainstorming, amazing partner, amazing at just thinking of different ways to organize things, analyze large data volumes, like I said, but you're not advocating responsibility. This is not a decision maker, a big, complex, strategic moments that we really need people's creativity for, certainly for now, probably indefinitely,

Robert Plank 11:30 very logical. So it's an organizer, right? And you think about when a lot of generative AI is used to, like, create content, right? Create text. And it's like, you give it all this stuff and have it like, write a summary, write a report, write a social media post. It's kind of just organizing the thoughts, ideas, paragraphs, sentences, stuff like that. So it's like, okay, it's it's not sentient, at least not yet, right? It can't think, but it can kind of rearrange these things in a way that makes sense. And so these are some fantastic ideas, yeah,

Rebecca Shaddix 11:59 can I jump in quickly? Or we got you're exactly right. So you can say, but it is just going to look at the volume of data it's been trained on from the internet, unless you're very specific about here are some post examples that I like. This is what I like about them, from this point of view, using this tone and these inputs right about this topic, ideally, with as many of the inputs as you can control for because otherwise you get this generic just kind of going to the mean of this average, because that's the most of what it's seen. And if you're not really careful with the engineering, this is the tone I'm going for, for this audience, for this input, for this intention, you just get a lot of generic fluff or average stuff. And that's what separates the effect of prompt engineering, in my mind,

Robert Plank 12:42 and it's easy to give up at that point, right? Say, Oh, well, this just doesn't work. This is this is garbage. And you're like, Well, you didn't give it the right instructions, like any computer since the beginning of time, right program it properly and and it kind of hurts my brain thinking about this when, especially, like, you ask the LLM to like, write the prompt for you, or you work with it to make the prompt that. Then you'll then reuse layer. It's like, ah, hurts my brain. But then, um, I noticed, like, when I have it try to help me, it does what you're talking about, right? It's, it puts like, in like HTML tags, like example, and then another HTML tag, like, slash, example, and in between there it says, like, hey, like, don't, don't make the output this. But here's an example of what I would like it to be. Because I know, like, early on, I was like, I'm afraid to give it examples, because I'm worried it'll just, like, use that or something similar. But it's like, okay, clearly say, here's the example, here's the inputs that I want, and so you you

Rebecca Shaddix 13:38 cannot copy it. But yeah, exactly. I mean, you're describing the typical hype cycle, right? We have some trigger moments, some inflection. We get into this peak of unreasonable expectations, that we get into the trough of disillusionment, and then we get into this plateau of productivity. We are right, right at the top, where people are expecting unreasonable things without the right constraints. I think we're going to head right now, as we have with the internet, with any other tech, into this trough of disillusionment, and then get to a reasonable adoption relatively soon. So yes, having reasonable expectations of what the tool is as a tool and not a replacement for a human brain is where it's effective.

Robert Plank 14:19 And you need that, that hype sucker, right? You need to get people using it widespreadly, mainstreamly, and then, and that way, then it'll kind of become part of what we do, and it's we're already on our way. And so I know that you, you're like, a Forbes contributor, and you talk a lot about AI. So as far as, like, your big ideas, is there any interesting idea that's been on your mind lately that you feel like sharing?

Rebecca Shaddix 14:44 big idea in terms of AI, or we mean big idea,

Robert Plank 14:48 yeah, in terms of AI, just like your hot take on things,

Rebecca Shaddix 14:53 my hot take AI is a tool, like any tool, just like the internet and getting disillusioned because it's not perfect. Right now, or living up to hype it wasn't designed for, is going to be short lived. I think that more effective data analysis and processing is here to stay. More effective tools are here to stay. And humans who know how to use it, who know how to engineer it effectively, know how to ask the right questions with the right inputs, are going to vastly outperform those who don't think it's relevant in their respective fields.

Robert Plank 15:26 Like any tool, need to use it wisely, right? They invent the automobile. Well, you need to learn how to drive it. You can't just look at it or play with it and crash it. You have to, like, use it in the proper way. That way you can get to where you want to go.

Rebecca Shaddix 15:39 Sure and attribute shortcomings effectively. Electricity, right was seen as less safe as candles initially because it wasn't set up effectively. But if we were to have just concluded electricity is not safe in homes because some of them caught fire early on because of poor installation, etc, that would have been the same sort of conclusion as drawing is asking something unrealistic now of AI and then concluding it'll never X, Y or Z. It's the implementation.

Robert Plank 16:09 Yeah, use it properly, and don't throw out the baby with the bathwater. And so as far as the future, like, is there anything that you're currently working on feel like announcing or anything that you're excited about in the coming months. In general,

Rebecca Shaddix 16:23 I am excited for the work we're doing at Gartner to see how we can continue understanding physician effectiveness using data and AI, understanding the impact that each provider really has on patient outcomes, which is great. I'm also excited to see how we get past the AI hype curve of just thinking, AI is the end all be all messaging, product strategy, etc, but being very intentional with the outputs that it's actually driving. And I think we're close that people who are intentional about the right application of technology for their tools, being mindful about asking, What can AI at this state, accelerate or amplify? Are creating great products, marketing them effectively, etc. So I'm excited to see this become more of a commonplace integration into our marketing and our tech cycle, just like we wouldn't be bragging that this is a web based application now, it's part of our integrated life and technology, and I think we're getting there pretty quickly.

Robert Plank 17:20 Yeah, we're still have a ways to go there, right? Because, like, there's so many websites or apps where there's, like, the star icon. I'm like, do you really need to have AI and everything will, I guess, at the moment, just to catch up. But soon, we'll kind of take it for granted in a good way, and it will be built in and so you mentioned about this idea of, kind of like judging or rating doctors. Like, do you think that might be misused? Because there's, like, the what's in China, there was the sesame credit where they kind of like ranked, like every person by number. Is there, like, a similar concern, like, if, if doctors are kind of like ranked in that way, or, like, have to do certain things to, like, stay you know, stay employed,

Rebecca Shaddix 18:02 your outputs are only as good as your inputs, and if a doctor has to, I don't know, play a system to stay employed. Certainly, it's a risk that is not at all the approach that we take at Garner looking really at their specific impact on patient outcomes, is aligned with everyone's incentives, because the way that medical billing often goes, the more interventions, the more tests, even if they don't yield better outcomes, they get billed as more productivity. And so when we look at the doctors that have the best impact on their patients, that's how we can start to understand the ones who are making the right decisions for their patients, and that's a great use. But of course, the outputs are only as good as the inputs. If we were to have a hyper controlled, inflexible approach to evaluating that wouldn't be effective, but we have over 550, specialty specific metrics across 82 clinical specialties to understand who was making the right diagnoses, who are making the prescriptions and prescribing the treatment best aligned to medical guidelines, and then whose patients are benefiting because of it. Those are all a great use of technology. But of course, the inputs are have to be reliable for the outputs to be actionable. And that's the mindful approach that we take

Robert Plank 19:15 very cool and like, and we hope that it goes in that direction, right? Like you always hear the platitudes about how kind of, like the holistic and treating the whole body is good, and the looking long term is good, and like doing testing and preventing instead of trying to treat something later is good. So we're hoping that AI kind of looks at that big picture and makes, you know, makes us all live longer, makes us all healthier, and as far as drawing attention to garner and what it does, I know it's kind of like a way outside of my space and way outside of my understanding, but is there a need to, like, explain what Garner does, and if some other, like medical company needs to, like, understand them, is there a easy explanation? Yeah.

Rebecca Shaddix 19:59 Sure. So Garner is a benefit that employers provide to their employees that offsets the cost of being seen by the best doctors in their area. Because these doctors are better, like we mentioned, they make more accurate diagnoses, they have the right treatments that they prescribe and better outcomes, and actually cost less to be treated by them. So employers are offsetting less effective, more expensive healthcare costs by incentivizing employees like you and me to go see better doctors. So in short, from our perspective, it would be covering the cost of seeing the best doctors, and then 75% of the patients who have Garner pay $0 in out of pocket medical expenses a year, because we completely offset the cost of their deductible most often. So ultimately, it's a benefit that employers offer to help get better quality care to their employees,

Robert Plank 20:51 and that's a very good thing. And how does some organization know if they're the right size or meet the proper requirements to contact you?

Rebecca Shaddix 21:01 Almost every employer is the right size. So some of the largest fortune, 50 companies use Garner and then down to smaller employers of just 50 employees. So we customize to the existing plan.

Robert Plank 21:14 Awesome. And so as we're writing down on conversation here, I know we're talking all over the map when that's okay. We I love the randomness, and I also like expanding my horizons and being curious and learning new things. But in the spirit of all that are talking about, like the medical field or AI or whatever, what interests you Is there a missing, hidden question in our conversation that you wish I would ask you, I just don't know what to ask you,

Rebecca Shaddix 21:44 let's see. I mean, not necessarily pertinent to the medical field, but we came in, we're thinking the most common go to market mistake that people make, so I think that that's something that I see pretty often. And I'd say that the most common mistake companies make with go to market strategy is twofold. One is not articulating a trip wire or what would need to happen by a certain point, certain metrics, certain results you need to see, at which point you'd pivot the strategy. But really just not articulating what kind of acceptable mistake there is because of the goals that you have. And I think of an acceptable mistake as the opposite of a risk. It's something that really doesn't matter materially to the results that you're trying to drive, but it could look nice. So it's often where we put resources into things like a launch or enablement or paid campaigns that just won't lose the needle. So articulating the acceptable mistake upfront and being mindful of what that is, is important to making sure that you're not putting resources into something that won't be as important to move the needle, because every company, no matter how big, has some constraint on resources. And so the speed to get something out, to the cost to the time, are the pretty common ones to think of as trade offs, how efficiently you can move versus how much time or resources you're putting into something, but that's where I see companies sometimes just do mediocre strategy because they haven't been intentional about what won't compromise their ability to get results.

Robert Plank 23:14 I like it, so be intentional of the speed, cost time. Trade off that way. You'll take that calculated risk, but you also will avoid the sunk cost. You kind of be have the right sizeness of your goals there. And so this is very fascinating and interesting to think about. We said at the beginning that you are at Rebecca Shattuck on LinkedIn, and it's always great when you make a media appearance, you a podcast, and then you get the right people finding you. So what reason would someone have to contact you if they found you on LinkedIn and you want them to send you a message and saying, Hey, I saw you on a podcast? For what reason would you like someone to be reaching out to you?

Rebecca Shaddix 23:55 I'm pretty frequently going to start posting more go to market, marketing advice, time management advice in general. So give me a follow, and I'd love to hear your thoughts or questions you have about those topics. And

Robert Plank 24:08 we could always use more help with that, right? Because it's not an easy problem to solve. And then maybe you get your immediate need or problem figured out, but then you grow as a person or as a company, or you have new problems. So it's always it's not a one time thing, right? It's like showering or brushing your teeth like you solve it for now, but you're gonna have to solve it again in a another day or so, or hopefully sooner and so, as people are checking you out on LinkedIn, that's Rebecca Shattuck, it's time to put you on the spot, to try to stump you a little bit, where I ask about a fun or interesting quote or lesson that has served you in life or in business. So what comes to mind as far as interesting quote or lesson?

Rebecca Shaddix 24:46 The quote that comes to mind I actually learned in disc golf initially, which is, know your miss, and that's used in disc golf to refer to if you often miss the left aim farther to the right, it'll feel off, but it'll probably. Get you closer to the target. And so even as you're looking to aim, you're like, Oh, this feels wrong. I feel like I'm aiming too far right. But if your miss is to the left, it's the right miss. I think in business and in marketing too We should know our own miss. Maybe we tend to be too analytical, analysis paralysis, looking for too much data before we move. So feeling like you're moving preemptively or too quickly is probably the right cadence. Could be the opposite too. If you tend to move too impulsively, too reactively, then slowing yourself down may feel like the Miss is too slow, but it's probably going to yield better outcomes.

Robert Plank 25:36 Ooh, that's, that's it hurts my brain to think about it, but in a good way, right? Just this. Need to have the self awareness and to experiment and to look at the variation and be like, Well, hey, if I you know, if I veer off this little bit, will I get better results? And we can't get that unless we have a team or have goals, or know our KPIs, and all those other things that we've heard of, and they're so important. But are they so important? Why aren't many of us implementing and actually living it? Well, that's why we need to follow Rebecca shacks on LinkedIn to get that kind of regular go to market and that time management advice, that way we can be on the right track. Whether we're working with AI, which we all should be right, it's kind of getting in everywhere, or if you're in education or HR or the medical field, check out Rebecca shattucks on LinkedIn for some of that regular help. So that way we can kind of get on track and build the company of our dreams and excel well in our career or our business. Can't do it alone. You can do with the help of Rebecca shattucks and her wonderful advice on social media. So thank you very much Rebecca for showing up, stopping by and giving us, like so many cool ideas to think about. Thank you.

Rebecca Shaddix 26:45 Thanks for having me. It's been fun. All right.