Podcast Episode 24
Navigating Data Strategy and AI in Healthcare
with Bryson Dunn, VP of Data Management at EXL
In this episode of Automating the Enterprise, join host Dayle Hall as he engages in a captivating discussion with Bryson Dunn, a seasoned expert in data strategy and healthcare. Tune in to gain valuable knowledge on navigating the intersection of data strategy and AI in the ever-evolving healthcare landscape.
Full Transcript
Dayle Hall:
Hi, and welcome to our podcast, Automating the Enterprise. I’m your host Dayle Hall, the CMO at SnapLogic. This podcast is designed to give organizations the insights, best practices on how to integrate, automate, and transform their enterprise. Our guest today is Bryson Dunn. Bryson is the healthcare vertical lead for data management at one of our strong partners, EXL. Bryson is a seasoned practitioner, mentor, trusted advisor within that team and in the industry in general, works with business development, data management, and a full-stack software engineering. Bryson, welcome to the podcast.
Bryson Dunn:
Thanks, Dayle, excited to be with you. I’ve listened to quite a bit of your podcast, but first-time podcaster, so very excited.
Dayle Hall:
Trust me, I’ve done a few of these and I wouldn’t say I’m an experienced podcaster. I think every single one is a new experience, but we’re glad to have you on the show. Okay, so we’re going to kick things off. I know we have a bunch of questions, a bunch of topics we’re going to cover. But I think the first thing we always try and do is let the audience know a little bit of your background. So I want to do two things with the background, if you don’t mind. One is your experience, your career, how you got to this type of role, and then a little bit more about how you ended up specifically on the healthcare side, because obviously health tech, this area, managing data, very specific considerations and rules and regulations. But let’s start with how you ended up in this type of role, and then let’s go into a little bit about how you ended up focusing on healthcare.
Bryson Dunn:
Yeah, thanks, Dayle. I spent my entire career in the data management space, so always on the consulting side, always on the implementation side, but started out with a data management software company, writing data pipelines, ETL, back in the day as a junior developer, and over the years expanded my expertise and my reach into other facets of data management, data quality, master data management, became an architect, and eventually made the leap over to the management side. And today, as you said, lead the healthcare vertical for the data management practice within EXL, which means I’m really responsible for everything, from growth to making sure that we are positioning the right solutions and offerings to our clients and ensuring we’re following through on our commitments and delivering effectively, making our clients successful.
How I really got into the healthcare space was one of the things I really appreciate about my experience consulting in my career is the opportunity to work with just a number of companies in the space and get a great breadth of experience, whether it’s within pharmaceuticals, or life sciences, or payers, providers, pharmacy benefits, brokers, even education, nursing education. It’s provided just a great, broad understanding of the space and the challenges that these companies have.
Dayle Hall:
I think that’s a really interesting point. Clearly, I’m not as experienced in this area, which is why I’m looking forward to the podcast. But I think sometimes when you think of healthcare, tech, data management, you think of HIPAA, you think of the more patient-doctor kind of relationships. But you just mentioned a bunch of things there, pharmaceuticals, education, those are the kinds of things that maybe not as popular or not as well known, let’s say, but that’s a massive arena. That’s quite the data management challenge across all those areas. Do you focus on specific areas more than others? With EXL, do you get a bunch of questions that could cover the gamut of everything? What kind of inbound or questions do you get in general?
Bryson Dunn:
Again, that’s one of the things I love about consulting is every day is a little bit different in terms of the questions that we get from our clients. But every one of our clients is coming at it from a different angle, depending on the role within the company, what they’re trying to achieve, and their company. As you can imagine, the concerns and motivations of a health insurance company are very different than a hospital system or an integrated delivery network. It certainly keeps us on our toes, and it’s very highly variable. I’ll say that.
Dayle Hall:
Yeah, I can imagine. Okay, so let’s dig in specifically around data management within healthcare, and you can give me some examples as we go. How do you see this area, data management specifically, solving challenges within healthcare that you’ve been involved with? How do you advise clients? I have this visceral reaction to the word transformation, digital transformation, because I think it’s used way too much. I think I’ve seen stats around 60% of digital transformation projects fail. And typically, I think that what that means is it’s because they’re either not defined or they lump a bunch of other challenges under the term transformation. But I would love to get your perspective. What do you see within healthcare, around this area? And are you really seeing transformation, or are we solving point-to-point challenges?
Bryson Dunn:
I completely agree with your feeling of the word transformation. I do think it’s overused and oftentimes a bit empty, right? So for us, when we look at solving for challenges in healthcare, our challenge becomes how do we make that real, how do we make transformation meaningful and valuable. And so for us, we’ve really been focused on defining the why of data. I think, historically, look back 5, 10 years ago, a lot of companies recognized the importance of data, but they were making the pitch of, oh, we need a data warehouse or a data lake because it’s a foundational capability, or we need a data catalog because it’s a foundational capability, just something we need. And right or wrong, I think a lot of business leaders that we’ve encountered could sometimes view those investments as just a large budget item that’s discretionary and not valuable, which I think the listeners of your podcast by and large will disagree with. And I’m in that camp for sure.
I think in healthcare, and really any industry, the challenge is how do you connect investments in data to business outcomes. So that’s really been our challenge. And in healthcare, what really excites me is that it’s so easy to draw a direct line between those things, from data to outcomes. And in healthcare, the outcomes are people’s well-being, people’s lives pretty directly, right, minds, bodies, wallets. You’re seeing things like using data to predict disease and improve preventative care, reducing costs, being proactive billing, and that sort of thing.
Sounds like table stakes, but I’ll tell you, and my wife just had a baby. And in advance, I called my insurance company. I said, how much is it going to cost to have this baby? And I could not get a straight answer. Even something as simple as that, that you think would be so easy, these are real problems that data can help to solve. And I’m not saying that- hey, there’s a lot of work to do. Let me put it that way. In the US, we spent twice as much as any other developed economy on healthcare, 16%. And we still have the highest rate of preventable hospitalizations and death. My viewpoint is data is a significant piece of solving that challenge and certainly something our clients are focused on.
Dayle Hall:
Yeah, you hit the nail on the head with the example. I remember being in that situation and getting the bill after my wife had our two kids, and still being stunned that we have good healthcare. I’m not complaining about that. We’re very lucky, we’re very fortunate. But still, when you look at that, not being able to give someone any kind of indication up front, that’s clearly a challenge we have to solve.
What I like about what you said, and I hear this with successful people in and around the IT space dealing with AI, or whatever it is, that if you start with business outcomes, or I think you and I have had these conversations, you start with the why. So why are we trying to solve this. And I think most projects will be more successful that way. And I’ve heard over these podcasts that if you don’t do that, you’re going to be one of that 60% of the digital transformation projects fail because you don’t really know what you’re expecting going into it.
I was actually on a marketing podcast, a marketing Zoom group this morning. And one of the questions to the CMOS was who’s being asked to do something in AI in their business. And I feel the example you said up front around move to the cloud, build a data lake, great, you know it’s going to have some impact, but if you don’t start with a business outcome, you don’t start with the why, and you’re going to fail. Give me some examples of what the why is for data and what are the things that you have to think about with your clients?
Bryson Dunn:
Yeah, this all really revolves around how we think about creating data strategies for our clients. And defining the why is a useful starting point because the things that you mentioned, migrating to the cloud, building a data warehouse, these are enablers. These aren’t outcomes. Integrating a new data source, retiring a legacy system, these are all good things that likely add value, but they’re not the outcome itself. So when we think about creating a data strategy, we, of course, look at it from that angle, from a data infrastructure and a data architecture perspective. Where are the inefficiencies? Where are the friction points? Where are the gaps? And how can we improve maturity?
But we really start with that top-down view of understanding what is the business trying to achieve? Are they trying to enter new markets? Are they actively looking to acquire competitors? Are they trying to move towards a more at-risk health delivery model, where we can help them, arm them with better data and insights at the point of care? All of these things are crucial inputs to where you invest your money and data. And you need to be able to track the outcome. That’s the important piece. It’s not enough to say, hey, these are the outcomes we’re driving towards, and here’s where we’re going to invest. You have to be able to prove you delivered.
Dayle Hall:
One of the things I try and do in the podcasts- because I know the people out there are going to be thinking about and how can they apply what you say to potentially their job. How many clients come to you with we know we want to move to a new cloud data provider, we want these outcomes? How many come to you with that already defined? And how many come to you and say, we know we have to do something, we know we’re looking for outcomes, and you help them get to the strategy? Because I do think that is an invaluable service, whether it’s healthcare or anything, someone, an advisor, consultant, group, even an internal panel or whatever, helping to get to the outcome. So is it 50-50, is it 70-30 that they know or that you have to guide them on that?
Bryson Dunn:
I would say probably 25% of the conversations we have are very high-level strategic like that. Frankly, many of our clients, the executives already know where they’re driving and tend to have a fairly good handle on this. And execution is really where they need the help. And this is a part of the variability that we were talking about earlier, right? Sometimes our clients come to us and they say, hey, listen, we just need additional capacity, we have a lot of data pipelines to build, or we’re retiring our legacy integration platform and migrating to something modern like SnapLogic, and we just need more horsepower. Or our clients would come to us, like you said, in a very key advisory capacity looking to us to say, listen, we know we can be using data better and driving more value with data, how can we do that?
Dayle Hall:
Yeah. I think the good thing is I think, obviously, you can handle both. But again, what I hear from these podcasts, people in your kind of position, or they are actually potentially in the client, their own client side, but they go into it looking for we know there’s 500 things we could do, but we’re going to chunk that into very discrete projects and deliverables, and it’s a journey. How do the clients react sometimes if you say, hey, this is great, it’s good you have this direction. You’re thinking about the outcome, so you’re on the right path. This is a five-year journey or whatever. Is that understood, that it’s not that quick fix? Or do you have to lay out as to why that happens with this kind of role?
Bryson Dunn:
Yeah, it’s generally understood. It’s generally understood that this is a long-term journey, not something that’s going to be fixed in a matter of weeks or months. And I think, really, the art of communicating a strategy to a lot of stakeholders within a company is to really communicate why it’s going to take so long, what we’re driving towards, and how you’re going to add value along the way incrementally. That’s crucial, because you can’t say, here’s your three- or five-year road map, and we’re not going to see any value until the end of it. So you need to be able to pick out, like you said, where are those quick win items that you can deliver on in two months, three months, very quick win, quick hit. And how do you sequence those so that they build on one another, and you can continually demonstrate value and you can demonstrate success, because if you go back in your back closet, you start building technology for too long without showing progress and showing value, people are going to begin to question those investments inevitably.
Dayle Hall:
Yeah, I like that. I like that thinking around the quick wins again. I think sometimes we lose track of that even if we have a strategy. Sometimes we’re thinking about that blue sky view and it’s going to be awesome. Again, like any business, or any function within the business, you got to show progress. If you’re building pipeline in marketing, where are we tracking for the year? If you’re closing deals, you’re expanding into a customer, What are the little deals? It’s all part of that process.
Do you have a specific example, something, a project you’ve worked on, maybe some data management project around healthcare that maybe has some governance controls? Do you have something that you can talk through how you managed it with the customer, the client?
Bryson Dunn:
Yeah, absolutely. I’ll give you an example of a recent project that we went live with. It was for a large group benefits brokerage. So this company, they sell all types of insurance, but they do also sell customized health plans for employers. And a key differentiator for them has been their ability to tailor plans and pricing, tailor to their clients’ unique employee population. That has a lot to do with the type of industry that they’re in, the type of workers that they employ, their social determinants of health, their health histories. And so all of that data is crunched and taken into account so that they can craft the lowest cost, highest coverage health plan for their clients and their clients’ employees.
So they’ve been using this data analytics solution to do this for quite some time, but the problem was that the data took about a month to refresh. It was hosted by a third party. And so they had just limited ability to scale and customize for their specific needs, but also to leverage the data for other purposes. And so they hired us to migrate the solution into the cloud and apply a major redesign with some key enhancements along the way. And it was a heavy lift. The solution ingested over 1,300 files from a wide variety of sources with linear formats and obviously had to be highly secure, just given the nature of health data. Yeah, always a challenge within the space.
And so we went live with an initial release of this that cut data latency from 30 days to five days and improved the speed of ingesting and onboarding new data by 15%. And we’re actively working on additional enhancements to refresh the data on a daily basis and further automate this solution just to reduce overhead and maintenance.
Dayle Hall:
That’s a pretty good project. I think if we were offering that kind of solution to our customer base, they’d be pretty happy with those kinds of numbers. Let’s move on a little bit. You mentioned itself, healthcare data is critical. Controls are critical. Let’s talk a little bit about not just the data but the metadata and why is that critical. Because I would imagine, again, it doesn’t really matter what type of company you are. The data and the metadata is important to make sure it’s actionable, findable, reusable, whatever. But I would imagine in healthcare, there’s some other more important controls there, or has to be done in the right way. So how important is metadata in these types of projects as well as actual the data itself?
Bryson Dunn:
Metadata is absolutely crucial. And this is such an important topic, Dayle. I try to harp on metadata every opportunity that I get with my clients, honestly, because it’s the secret sauce to so many things, whether you’re talking about automation or security and controls, or you’re talking about AI. Metadata is really the fuel behind all of that, i.f you ask me.
Dayle Hall:
Explain why. Tell us why.
Bryson Dunn:
Yeah. I relate it back to the example that I just gave of that last client that we delivered the project for the cloud migration. If we were to go out and hand code 1,300 different data pipelines for that client to move the data into the cloud, it would have taken us years, so it’s just not tenable. And so instead, what we focused on was the metadata, focusing on where is the data coming from, how do we connect to it, how is it structured, how does the data need to move and transform into our target data model, and what are the quality checks that we need to apply along the way to ensure that the data is valid and correct.
With that metadata in place, we can then implement a few dozen higher-level jobs, data pipelines to move all of that data dynamically in a configuration-driven way. And that’s just one example. That’s just an automation example where you can drive incredible efficiency with just some basic metadata about how data is structured, where it is, how it needs to be transformed, the rules that need to be applied to it. And in healthcare, it’s absolutely crucial. I think, in particular, to know how data is moving within your infrastructure, where it resides, who has access to it within healthcare is just absolutely critical.
Dayle Hall:
I’m interested just on that topic. But again, when you’re out there talking to some of the clients on the consultant piece, is there ever a disagreement or a challenge that you have to remind people, the clients, around why metadata is so important? The reason I ask is I love what you said, which is metadata is the secret sauce, whether it’s automation, security, using it in AI, making sure the models are fair, or ethically, we can go down that path on a completely different podcast. But do you ever have to remind the client why this is the most critical part? And do they immediately get it once you’ve explained it? I’m sure they do. But do they see that as the most important part? Or are they thinking about other things?
Bryson Dunn:
Often they’re thinking about other things, frankly. A lot of our more enlightened clients will recognize it. The truth of the matter is it takes more time. If you’re onboarding a new data source or building a new data platform, migrating it to the cloud, it takes time to put that metadata in place. It doesn’t happen magically. Now there are tools to automate some of that. But there’s always this last mile that’s needed.
Oftentimes, people are just very focused on delivering the project at hand that you will see metadata being put on the backburner, being deprioritized, because we just need to build this, get it shipped, get it into production. And listen, sometimes, frankly, that’s the right choice. I’m not going to say that every client all the time needs to prioritize metadata. It’s always a trade-off, it’s always a balance. But I think it would be a mistake for any organization these days to not be considering metadata as a key component of their long-term data capabilities. And that will really support and enhance all of their other initiatives.
Dayle Hall:
Yeah. No, that’s interesting. Because one of the other terms we talked a little bit up front, digital transformation, the thing I hear a lot about is this term data democratization. When you hear that term, what does it mean to you? And how does that relate to metadata versus the other core of the data that’s been moved around systems?
Bryson Dunn:
That’s another one of those terms that can ring a bit empty, frankly, unless, again, you put some meat to the bone there of data democratization, because a lot of people talk about that. So to me, what that means is arming your business users with access to high-quality data assets that are timely, quality, understood, and complete. That’s really data democratization. Some people, I think, even before data democratization was a term, they would call that self-service, self-service, self-service bi. The idea has been around for a while.
But to me, that falls a bit short of where we need to be thinking and how many of our clients are thinking. And that is, instead of just having us build it, and they will come model, hey, here’s the data, go after it sort of mentality, we need to be more proactive in how we’re embedding data and insights into workflows,into business processes, and making it not optional. And so being just more deliberate on how you’re doing that. And that takes time. That takes a lot of change management and a lot of integration. But I think that’s where we really need to be.
Dayle Hall:
Yeah. For the listeners, I can’t say I didn’t smile when you said a lot of integration because, clearly, that’s important to us at SnapLogic. So that’s a good thing.
Bryson Dunn:
Trying to throw you a bone there, Dayle.
Dayle Hall:
Yeah, thanks. I appreciate it. My CEO gets upset if I at least don’t throw in one SnapLogic reference. So we’ve got that. So let’s move forward. Let’s actually talk about- we hit on it a little bit, which is actually defining the strategy. And sometimes you get clients that are already further down the path and you can help them refine, help them on outcomes, which I think is good. But let’s say someone comes to you and said, look, we know we have to do something, we know we want to move. We’ve got some on-prem, we want to move to the cloud. But we don’t really have as much of a strategy today. If someone’s out there and they are thinking about that right now, what are the key things that you, EXL would sit down and say, okay, here’s where we start, here’s how we start to- before we build anything, before you go and buy a cloud data warehouse or an integration tool, here’s where you start.
Bryson Dunn:
Yeah, no doubt. Every consultant in existence today has a basic recipe for this, and we’re no different. First, you got to define where you are today and the realities of how you’re operating as a company today. Then you want to define where you want to be, and then you put that plan in place to connect those two things. That becomes your road map. And like I said, we’re no different than any other consultancy, and that’s our basic recipe. I will say that we do focus up front keenly on that business value. We were talking about that earlier. How do you define it up front? What are those outcomes? How are you going to baseline? How are you going to measure the value? How are you going to baseline the value and prove the value after the fact?
That’s really what we would advise our customers to do. And it’s an art. It’s not a science, no doubt, because you need to be engaging with executives across the organization who have sometimes very different priorities, objectives, challenges that they’re trying to solve for. And politics is always a factor. That’s just the reality. And so coming up with that unified view at the enterprise level of what makes sense to do first, second, third, what are those business priorities, and where is the value, that can be a challenging thing. So I will say that it is incredibly important to get those perspectives. Otherwise, there will be pockets of the organization that struggle with adoption, struggle to support the program. So having that buy-in is crucial.
Dayle Hall:
Yeah, I’m going to ask you in a second about the impact of this with some of the innovations around AI, which are obviously important. Getting people involved is what you just talked about, making sure that it’s understood across the enterprise. How do you do that? And I asked that question because, again, if I’m out there, and someone’s listening to this and they’re embarking on this kind of project, or they’re thinking about it, I honestly believe that one of the challenges is getting the enterprise aligned to what’s going to happen. They don’t care what industry is.
And what I like too, what you said earlier is about providing data to the business users, which I thought was critical. It’s not just about what IT you’re trying to do to support the businesses. It’s the business users having access to the data. How do you guide organizations to bring in the right people at the right time? You don’t have to go into a full change management plan. But how do you guide them to make sure that essentially the strategy and whenever you kick off is aligned from day one?
Bryson Dunn:
Yeah, it’s a challenge, no doubt. And it’s interesting. You have to take a bit of a different approach, depending on who you’re talking to. And it’s often useful to start with IT because they’re invested in making data and technology successful. So there’s really a lot of allies in that space. And when you speak to IT, they’ll have a really good idea of where some of the issues exist and what can be done to make things more efficient and deliver things more effectively. So it’s a good starting point for the conversation. And then it kind of arms you with the information that you need to go and have those conversations with some of the leaders on the business side. And you can’t get stuck in analysis paralysis mode either.
So when you go and you have the conversations with the business, you can be talking to hundreds of people all year long, trying to get every perspective. But you really need to focus the conversations on, number one, the corporate strategy from the C level and what are they trying to achieve. And then drill down into that a couple of levels in terms of the leadership within the company to understand what are they really trying to solve for within their company. Is it to make a process more efficient? Is it to integrate a new data set so that they have better insight into new treatments and new medicines, whatever that looks like. So you take all of that as input and share those perspectives with the other leaders. Sometimes the culture of a company is one where a consensus is mandatory. And in those circumstances, that sort of collaboration and ideation in a group setting can be very productive. Other times you need to get people one on one. So it’s important to tailor your approach based on the company and their culture.
Dayle Hall:
So I mentioned it, I’m not going to delve deep into it. But obviously, the recent announcements around AI and these generative AI tools are obviously built on data. I’m not going to go into that in too much detail. But are you now starting to see some of those questions when you get to what we were just talking about, which is our data strategy and building that out and helping clients to define it? Are those questions now front and center? Are they a key part of it and you have to be prepared to also say, here’s where you need to look at it today, and here’s where you need to look for the future?
Bryson Dunn:
No doubt. I was making this comment to my boss just this morning. I think every board of directors across the country today is talking about generative AI, ChatGPT, these days. It’s everywhere, a lot of hype, but there’s a lot of disruption and a lot of reality underneath all that hype as well. And two of my clients just this week have asked how we can help to enable implementation of large language models, generative AI within their companies. And so we’re definitely getting the question. It’s definitely top of mind for our clients. No doubt, it’s going to be impactful for their business and disruptive for their business, just about every business. It’s going to make a big splash.
And the long and short of it is you need to have good data to feed to your model. Otherwise, your model is not going to be very accurate or effective. And again, I don’t think that’s going to come as a surprise to anybody, any of your listeners of this podcast. When I was saying that metadata is really the road map or the secret sauce for AI earlier, it really is. You can think of it like a deep learning lesson plan. When you go to train your model, you point it to all of these different data assets. And all of a sudden, the AI understands what the data means. Is the data accurate? Where the data come from? When was data last refreshed? And all of this is going to be crucial information to make sure you have an accurate model. Because I think it’s become obvious over the last weeks or months that it’s not going to be enough to just drop an API to ChatGPT into your CRM and call it a day. You need to have a model that’s trained on your data, understands your business. So the question becomes how do you train it effectively? And data is the answer.
Dayle Hall:
Yeah, I love that. And you come out with two zingers now. I’m just going to let you know. One was metadata is the secret sauce, love that, and a deep learning lesson plan. I like that. This is going to be great. We just got so many sound bites for this one. I love it. So you start the strategy. There’s obviously questions that are coming up. One of the things that I think is also critical, and you talked about this a little bit, which is the milestones on the bigger journey and getting the wins and showing ROI as you progress. How do you know, what are the kinds of things that you set up to say, we know our strategy is being successful when it does whatever it is? Again, do you have some examples of how you build those out? What are those ROI things that you build into the project? So the strategy, that isn’t just, hey, in five years’ time, this is what it would look like, but you won’t see anything until then.
Bryson Dunn:
Yeah, exactly. The short answer to that, Dayle, is you know that you have an effective data strategy if you can point to the measurable value that you’ve added. And the business agrees that you’ve added that value with the data investments. So until you can do that, you can’t claim to have an effective data strategy.
The other point that I’ll make here as well that I think is often overlooked is the process of creating a strategy, any strategy, not just the data strategy, shouldn’t be viewed as a onetime activity. I think it constantly needs to be revisited to say, okay, here’s the road map that we put in place last year or even six months ago. How are we tracking to that? Have we added that initial value that we thought? And if our success has been mixed, why? Were there challenges we didn’t foresee? Have some of our underlying assumptions changed? Has the economic environment changed? Whatever that may be, you need to continuously revisit, reevaluate your strategy and incorporate those learnings. When you talk about delivering value iteratively and incrementally, I think as important as that is, I think it’s also just as important to be nimble and adaptable to pivot when you need to.
Dayle Hall:
Again, you’ve got more experience with these large type of projects and helping to build those strategies. So I think that’s a good way of looking at how you build these kinds of things. We just mentioned around AI and the importance, but I think I want to just finish with just a couple of short questions around AI specifically, because in my arena for marketing, even in our industry with SnapLogic and who we compete with, I feel like we’ve been AI washing for 10 years, now we’re generative AI washing. There is a little bit of this is going to solve- all the things that we couldn’t actually solve for before, we can now solve because of this.
And look, I have to tell you, I am more excited about some of the things that we’re doing with SnapGPT built off these generative AI tools than I’ve ever been on the product. But there’s also this expectation that it is a panacea. I think we talked beforehand, this is something that’s good. That’s a feeling like we can solve all these problems with that. So what do you think around some of the key areas that it can be leveraged today? And how do you respond when you know someone, a client or potentially when you’re talking to anyone that says, I think this is going to solve X, Y, and Z? How do you make sure that we’re doing it the right way? And we’re not even talking about the ethical AI part of it, but just running it in our business, what we can leverage and what it can solve for today?
Bryson Dunn:
Yeah. I think that, like we were talking about earlier, there’s a lot of hype, no doubt. My sense is that irrespective of the hype, this is going to be likely the largest disruption that I’ve experienced in my career in terms of technology. If I was smarter, I would have used ChatGPT to write me a script for this podcast, and I didn’t. But it’s very well capable of it. And use cases like that are very within the realm of possibilities for generative AI today. If you talk about, for example, a chemo patient that has dietary restrictions and needs, you could use generative AI to draft a meal plan for that patient.
Now it’s never going to be- at least not today, it’s not going to be generate the meal plan and deliver it to the patient. I’s going to be an accelerator for doctors, a starting point that they can use to refine before finalizing and putting into place. The same sort of thing in terms of a care plan for a diabetic. These are all very real things that generative AI can do today.
Now there’s a lot of room to grow. And there are limitations to the current instantiation of generative AI. And I would point to the key limitation being you have to ask it for an answer today. You have to prompt it. There’s a lot of conversation about how do you optimize and craft the prompts that you send to these tools. And right now, that’s a key limitation and how you interact with them. Long term, I think it’ll be more proactive. It will pop up with insights on your dashboard, on the applications you use day in and day out proactively. I think that will make it much more useful. But ultimately, the information that it delivers to you needs to be grounded in good data, good insights. And so when you talk about what companies need to be doing today to prepare for that, it’s invest in those data assets. And it doesn’t have to be so that you can use it one day for AI, these investments can add value now and they’ll continue to pay dividends along the way.
Dayle Hall:
Yeah, I like that. I like thinking about it that way. And I think, internally, as we’re talking about it, I don’t know if you’ve ever met Gaurav, our CEO, but he’s like a professor. He’s very well read. And he refers to this as like the second Socratic method. The first Socratic method is how do we get the answers, how do we know the answers that will get you ahead. He refers to this now with generative AI is how can you ask the right questions to get ahead? Because the answers are there now somewhere. But how can you figure out the right questions to ask, not just understanding the answers, which, again, he’s more of a professor than I am. But I thought that was an interesting way. And I think this could help on things like data strategy, or frankly, anything to do with data, which is if you can ask the right question, you probably will get the answers that you’re looking for. But there’s still I know we’re just scratching the surface on this.
Bryson Dunn:
Yeah, no doubt. I love the way of thinking about that, though, the very philosophical way of how you ask the question. I think the other- this is not new. There’s a lot of discussion around this, because how can you trust the answer? Can you trust the answer?
Dayle Hall:
That will give me a whole podcast on that specifically.
Bryson Dunn:
Exactly. Yeah. Yeah. And that’s a consideration of a lot of our clients as well, especially in highly regulated markets, like healthcare, is the risk of providing wrong information to patients, the damage that can be done is quite high. And so minimizing that, especially in the healthcare context, is absolutely top of mind for our clients.
Dayle Hall:
Yeah. What do you say to people now just in terms of preparing for it? So obviously, we talked about some specifics, and it’s not going to solve everything. Is there a feeling now when you talk to clients that they’re like, oh, this means we can go faster with everything? Or how do you temper that excitement and potential? Actually, I don’t know in healthcare because of the controls on data. Does it feel like it’s still we got to do this now, or is it slightly different in healthcare?
Bryson Dunn:
Yeah, the expectations are tempered. Within healthcare, generally, I’ll say that they are more risk averse, just because it’s a highly regulated space. That’s really when I speak more towards the payers, the providers, you want to talk about medical supply companies and manufacturing companies and related fields, they’re absolutely trying to move fast. And it’s this philosophical debate between the business who says, wow, look at these tools, and I can use them today and they’re so powerful. And perhaps the folks who are more well versed in data and analytics understand that you need to put appropriate governance and guardrails in place to mitigate the risk of, like we’re talking about, incorrect information being provided, or data being exposed inappropriately. It is tempting to try to move fast and leverage the technologies that are all of a sudden, just in the last few months available, but it’s a balance.
Dayle Hall:
Yeah, for sure. Well, it’s been a great podcast. I have one final question for you as we wrap up. It’s a little bit slightly off topic, but it does relate to AI or generative AI or ChatGPT. If there’s something that you could think of that is exciting for you, something that is going to be that what we’re seeing on AI is going to lead to major changes, could be personal life, could be healthcare specific, could just be IT, what’s the one thing that you’re looking forward to that would actually, you think, would improve our situations, not potentially cause more challenges around this specific area?
Bryson Dunn:
Yeah, it’s a great question. I think the number one thing has got to be the promise that AI has to improve proactive preventative care. So much of our health issues in the US, in particular, they’re preventable. And I think AI has a real ability to increase the bandwidth of our specialists and our doctors to be able to detect things earlier that may have been missed. And so for me, that’s what I really get excited about.
Dayle Hall:
Yeah. All of us at some point come into contact with the system that you’re helping to guide and support. So I think that’s a perfect end to the podcast. But I’m hoping that when we do the promotion for this, we’re going to keep that metadata as a secret sauce because it’s one of my favorite quotes I’ve heard. So Bryson, thank you so much for being part of the podcast.
Bryson Dunn:
That was a lot of fun. Thanks for having me, Dayle.
Dayle Hall:
Right, Thanks everyone out there for listening. We’ll see you on the next episode of automating the enterprise.