Pharma Sessions
Pharma Sessions
Building AI Workflows That Actually Work in Pharma with Adam Mico
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
Pharma Sessions is hosted by Jonathan Kaskey
Follow along on LinkedIn: https://www.linkedin.com/in/jonkaskey/
Or join the Pharma Sessions Substack:
www.pharmasessions.substack.com
You have to be more than skillful in one tool. So if you're practicing, if your main job is data visualization, you shouldn't be just focused on one specific tool. You should be able to learn how to incorporate that in multiple tools because you need to meet businesses where they're at. Many businesses aren't at just one tool. There may be multiple tools, especially in pharma. You may have Spotfire, you may have Excel, you may have Smartsheets, you may have Power BI, you may have Tableau, and on and on and on. So it goes on infinitely, but the more you know, the more valuable you make yourself with the company and you're able to establish value because you have this very generalist skill set and not a specialist skill set. And I think generalist skill set is kind of where we're gonna need to go going forward.
SPEAKER_01Hello, hello, and welcome to Pharma Sessions, a place for pharmaceutical leaders to come and learn from each other. I'm your host, Jonathan Kaskey. Technology and market trends are bringing change at an ever accelerating rate, and no person, team, or company can afford to be left behind. Here, we dive into the strategies and tactics that our guests use to tackle these challenges and create new opportunities, and how you can do the same in your own organization. On today's episode of Pharma Sessions, I'm excited to welcome Adam Miko, principal data and analytics strategist at Moderna, and one of the most hands-on practitioners I've met when it comes to integrating generative AI into real enterprise workflows. Adam's journey into analytics is anything but traditional, and that perspective has shaped how he approaches data visualization and now AI, not as abstract capabilities, but as tools to remove friction, unblock teams, and help people do better at work. In this episode, we're diving into how custom GPTs are actually being built and used inside pharma organizations, what analytics looks like in a world increasingly shaped by AI, and how data visualization, text-based insights, and enablement all come together inside modern workflows. AI, AI, AI seems to be the theme of this one. So with that, welcome to the show, Adam.
SPEAKER_00Thank you so much for having me. It's an honor to be here, Jonathan.
SPEAKER_01No, I'm very pleased to do it. And okay, so before we get into all the work stuff, I always like to start with a little bit of an icebreaker. And we're recording this probably a day after a major snowstorm has, you know, paralyzed half the country. So it's about 3:30 in the afternoon. Tell me what have you had to eat so far today?
SPEAKER_00So, what I had to eat so far was uh Thai food. I live with my wife and my mother-in-law and my dog, and my mother-in-law is probably the best Thai cook I know. So, fortunately, had some incredible Thai food both for breakfast and lunch. So, uh completely spoiled. That's amazing. It warms the soul.
SPEAKER_01That's yeah, I'll we'll record the sequel in person. I'll uh I'll be there uh in your living room for the next one. All right. So I kind of alluded to this in the intro, but your path into data and analytics wasn't traditional path. And now you've really become a recognized leader in visualization and generative AI. Can you walk us through a bit how that journey unfolded and what pulled you into the space?
SPEAKER_00Yeah, my journey is very different, as you mentioned before. I mean, I didn't really start the data and analytics journey until after I was 40. I had a whole career in unemployment insurance law. So I was working in the public sector. And when I was introduced to Tableau, which is a data visualization tool, I was kind of known as the guy around the office that was good with Excel. So they're like, You're a trainer, you have a little bit of bandwidth, maybe you could play around with this and see what we can come up with. And that just sparked a love for data and analytics beyond what I could ever do at Excel. And at that point, it was basically working in a silo um with my employer for about six years before I did anything publicly. I was very quiet. But then I left my career as a trainer of unemployment insurance law and a fraud investigator to work as a data analyst and a business automation specialist. And then I was promoted a reclass as top as I can get within a couple of years. I'm like, what else is there for me to do? So I wanted some motivation. So I wanted the opportunity to give back. And that's what I thought I was doing when I started doing it publicly, but I learned I had so much to learn.
SPEAKER_01And that second role after leaving the unemployment insurance company, was that in Pharma or was that somewhere else?
SPEAKER_00Completely somewhere. I worked in the public sector for the state of Wisconsin. I was working through the beginning of the pandemic. So a lot of Saturdays I was working trying to get reports to the governor's office and so forth, then unemployment insurance backlog as a data analyst and also projections.
SPEAKER_01Wow, that's fascinating. So you joined Moderna after the pandemic. That's that must have been a really at the tail end. So that must have been that crazy time to join.
SPEAKER_00So I left the public sector. I went and then I worked as a tableau evangelist for a global consultancy for about seven or eight months. And then they're starting their tableau practice at Moderna, and I was recruited for a technical lead for that role when I joined. And that was four years ago. I just had my anniversary on January 18th.
SPEAKER_01Congratulations. So even in those four years, there's been a tremendous amount of change, right? And a lot of people in analytics, I mean, I talk to them every day, and there's existential moment of what was our role? What is our role with AI? From your perspective, how do you see the role of analytics changing as generative AI becomes more embedded in everyday workflows?
SPEAKER_00Great question, Jonathan. And one of the things I was thinking about early is I've always tried to think about make uh future-proofing myself and future-proofing my skill set. So I started to see the changes and the evolution that what's possible with generative AI as soon as it kind of came to the picture. So I started thinking very deeply about that. And I wrote a couple of blog articles about that. Primarily, I came to the conclusion that you need a plus one skill set. In the past, if you're a tableau developer, you could focus a lot on the front end stuff. But with generative AI, more people will have the opportunity to upskill quicker, and more visualization development will probably be expected of just business people that aren't developers because you won't need to learn how to code as much or anything like that. So if you're a Tableau specialist, you need to figure out what's my plus one skill set? Is it product management? Is it data engineering? Is it generative AI or AI? Is it machine learning, which would be associated with AI or whatnot? So those are important factors that think is like how do you diversify your skill set so you don't get locked out?
SPEAKER_01That makes a lot of sense. And there was a actually one of our clients, so we have own version of like a Tableau, like a proprietary business visualization thing. And they came over from Power BI, right? Power BI and Tableau being the biggest ones. And it wasn't necessarily anything to do, well, I shouldn't say it wasn't anything to do with the technology. They were saying essentially what you were saying, but they had a kind of a different spin on it, which is we need to really get very clear at elevating our value to the organization. And if the value of building reports is going down, then what is the value that we can bring? And what they aligned on was we need to really focus on delivering strategic insights. That's kind of the purpose of all of this data visualization is strategic insights. And they're actually their team is growing. It's it's not shrinking with all of this. And I think they're very happy to be kind of elevating their position. So I think that that's part of it too, is just like how we talk about ourselves in analytics needs to be reflective of the value that's being delivered.
SPEAKER_00I think that's a great spin. And you've touched on another point, is that you have to be more than skillful in one tool. So if you're practicing, if your main job is data visualization, you shouldn't be just focused on one specific tool. You should be able to learn how to incorporate that in multiple tools because you need to meet businesses where they're at. Many businesses aren't at just one tool. There may be multiple tools, especially in pharma. You may have Spotfire, you may have Excel, you may have Smartsheets, you may have Power BI, you may have Tableau, and on and on and on. So it goes on infinitely, but the more you know, the more valuable you make yourself with the company and you're able to establish value because you have this very generalist skill set and not a specialist skill set. And I think generalist skill set is kind of where we're gonna need to go going forward.
SPEAKER_01And now another skill set that you've developed, and what's interesting to me is you're kind of doing this publicly is developing custom GPTs. So talk to me a bit about that just at the beginning. Like what made you think, hey, this is I can do this or I'd be good at this, I want to do this. And then what was around your decision to do this in kind of more of a public manner than strictly behind the scenes?
SPEAKER_00When I was introduced to Tableau over a decade ago, it was this like wow moment for me. It's like, this is mind-blowing to me. This is something I could utilize. I had that same moment when I was introduced to generative AI. We were able to get it very early. So I was developing custom GPTs internally early, but I learned very quickly that if you only build internally, your skills don't develop as quickly. So I created completely different versions of custom GPTs that I would make it work because obviously I'm not going to include anything. And I rebuilt them from scratch. And what I do with them is first of all, it helps me understand everything I need to know about building a GPT and what works and what doesn't work. And two, I could kind of crowdsource that feedback to improve my internal GPTs. The GPTs aren't identical, but they're close enough where if feedback applies to one, it could apply to another internally that you can make it better, where people can really get the value of what you're putting together. So, yeah, don't make identical GPTs as you would in work, but make it similar enough that if you do get feedback, that you can improve what you do at work. And it's a passion of mine too. So developing publicly is like, I want to incorporate tools that others could use as freely as possible. I don't like to be the gatekeeper of who gets in and who doesn't get in in data analytics or anything related to uh technology. And if I have the skill set to do that, it's my passion to bring some of that out and give back because I was fortunate enough to get plenty. So I want to give back as much as possible. And that's one of my main goals is here are tools that you could use that can make your life so much easier, and they're free. I'm not charging a penny for them.
SPEAKER_01That's so interesting and it's admirable. And so, can we just dive a little deeper on that? So, you talk about crowdsourcing feedback. What types of are these external usage patterns? Are these ratings and comments? Are these actual metrics about where people are going within the GPTs? What's available and what might you get from a learning perspective from having a whole bunch of people on it that you wouldn't if you were strictly designing in a closet in your Moderna office?
SPEAKER_00Exactly. So one thing is that you get anonymous feedback, so people could send you feedback, and then when you're writing a custom GPT and chat GPT, you could do it anonymously. So you get that anonymous feedback and also share a lot on LinkedIn as well. So if people have feedback, and I've always been had an open door policy, if you have feedback constructive or positive, open up and share how you're using it and how it could be better. And that's what it's so a combination of both. And there are ways that there are other ideas when you crowdsource that you wouldn't think of if you're working in closet or a silo, if you will, because people use it and like this is good, but then other people that are practitioners and or experts in the field that you're building GPTs on, they could have other ideas that you wouldn't come up with on your own, and that's how you can make both those GPTs better and the internal ones better.
SPEAKER_01I'm just really curious about this. So, how long did it take you to build your first one? And how long does it take you to build one now?
SPEAKER_00That's a great question. Uh, so when I started building GPTs, it was kind of interesting because how my brain works, I'm autistic, so I'm a systems first thinker automatically, and that kind of works with how GPTs uh line program when they're thinking their systems first uh structure, even more so now with the new models than it was before. But also I'm a very structured person. So a lot of how I think mirrors how you would uh build instructions into a GPT, so it came naturally to me, but my concepts weren't there because I wasn't really understanding what was capable of it. So my first GPTs were novelty GPTs, like I want to do a GPT with this personality. So I think one of my first public ones, and this is kind of funny. If you ever heard the good place, the television show, and there's all these versions of Janet. So I made a GPT with different Janet personalities, including Disco Janet, Evil Janet, Janet Janet, and all those, uh, just to see what it could do. It's a very novelty. I stand corrected. The first one was Ted Danson's birthday, because that's kind of a meme that was on Twitter that every day it was just like programmed only to say it's Ted Danson's birthday when it was. So these were real novelty ones. So they didn't really take me long to do, but I didn't understand what the capabilities were. So the only time I really started to build and understand those capabilities when I was figuring out things is like there's this interesting use case. How do I use it for that? And the ones that were more complex that took me longer to build initially were product assisted ones. So I have a public one called Tableau Virtuoso. So it includes a product knowledge of Tableau, but you have to create instructions on when to incorporate the knowledge and so forth. So understanding how to do that took me a couple weeks to get that one down. And then I created a couple other ones. One was this critique pro, which is you do a screenshot of a dashboard and it does automatic scoring for you as soon as you publish it. That took me several weeks because the scoring waiting at that point, and that it was you really had to create defined outputs for the GPT based on the model back then. So you had to do a lot of rules, you had to test the weighting, see if the scoring was right and so forth. So there's a lot to go into it. So most of the time was I do the initial iteration and I learned quickly that you write in markdown code, which is kind of a structured text because the GPT understands that more if you're typing in regularly. The other thing I learned very quickly is that when you're configuring a GPT, there's a little section on Chat GPT that has this really automatic prompt to write what your GPT wants to do, and it has this configure that's useless. So that doesn't do a lot. So there's a lot of learning and trial error based on testing. It's like, you know, maybe I can make this quick GPT on the fly by using configure GPT. No, it doesn't work at all, it's horrible. But then you have to test and test and test. But just recently, and this is another funny thing, this is a little bit of a tangent, but in April, I wrote an article on designing custom GPTs. It's been by far my most popular article on Medium. I think last tech was about 17,000 views, and that was, I believe, April 30th. And with the new model coming out initially in November 17th, and the most recent 5.2 in December, everything changed. So it became obsolete. That what I wrote in April about custom building GPTs became completely obsolete by December. So I had to write a guide to custom GPTs part two because so much became obsolete. So it's interesting how you're always having it evolved. But the great thing is that the models' capabilities are getting much smarter. So instead of writing strict rules and making strict guidelines for GPT, you build in workflows. Instead of pouring everything in one knowledge document, you make smaller knowledge documents and create a knowledge index so it knows exactly where to go, so it works efficiently and doesn't hallucinate. So there are many things that the new model does, so much more better than the original one, but then there's a lot of rework that needs to be done. All those old models you're built on 4.0, you need to change them because they don't really work well. They're brittle under the new models.
SPEAKER_01It's so fascinating to be, and I would just encourage people, if you're listening, you could also take a middle step before you are developing these yourself. So, what I mean by that is so on my team, this is something that happens all the time in software companies, is product will come to sales and they'll say, Hey, what are people interested in? What are they talking about? And it used to be that we would have a call or have a conversation or try to describe something that we had in mind, and then they'd say, Okay, is that feasible? Is that not feasible? What do we need? What my team has started doing now is I think they're using mainly this platform called Lovable, and they just like talk with it and come back to product with wireframes. And somewhat clickable, doesn't really work, but it's not meant to work, but you can kind of explore around and see how what the design the person had in mind. And that's been allowing our dev team, it's almost like an improv where they always yes and right, like they can take your wireframes, they can yes and it, and then they can turn it into something really awesome. But our first several iterations are back and forth, are all via these low-code, no-code wireframing type things. And that has just allowed us to move so, so much faster with development because we're literally reducing timelines by weeks and months just to understand what each other is even talking about. It's really, really cool.
SPEAKER_00And you're not going down those negative paths either, because when you have a good contextual wireframe that both the business and tech understands, they can communicate. Many times, if you go to a development team, the requirements are vague or not so clear to them, or they get something that's in a storyboard that's looks like somebody chicken scratched it. They're building based on what they think the customer wants or the business wants, as opposed to the business being able to present something, speaking tech's language and tech being able to build to that. It's way more efficient. You don't go off those beaten paths where you go down this road very long ways, halfway across the country, and realize I should have turned the corner back in Philadelphia, as opposed to being in the Midwest at some point going down that road. It's like I'm gonna have to not only start over, I'm gonna have to do a lot of rework because a lot of things broke along the way. So the time and savings that are implied with doing these better storyboards or better work for the tech team is gonna be super helpful for everyone. It increases the trust too with the business as well, because the developers can trust the business and don't go off on their own to engines.
SPEAKER_01If you've been enjoying the conversations here on Pharma Sessions, you should know they're made possible by the team at Excent. XSunt helps life sciences companies turn complex data into clear, actionable insight. For years, Xunt has made complicated data sets simple to help commercial, medical, and operations teams map what's happening, predict what's next, and make stronger decisions faster. And now there's an added AI layer that makes everything work so much better. I was actually pretty jaded about some of the AI approaches I'd seen, but when XSunt showed me theirs, I actually left my job to come work for them. It's really awesome. So if you want to understand your market, your customers, or your performance with more clarity than ever, check out xunt.com. That's xs-unt.com. All right, let's jump back into the episode. And I think it was how this got introduced is somebody on the team just did it. They didn't like asking, hey, can we do this? Is this a good idea? They just kind of did it. And I think that was another really interesting thing was like uh, I don't know, just taking some initiative internally at an organization and bringing us to the future whether we wanted it or not. Almost. But it worked out great. So let's bring it back a little bit to the data visualization piece, right? So a lot of your work sits at this intersection of data visualization and generative AI. How do you think about when insights should be visual versus when they should be delivered through text or conversational interfaces?
SPEAKER_00A lot of times, uh people just need a quick overview of what's going on before they drill down. So I always look at it as like people get overwhelmed very quickly. There's so many things going back and forth. If you could visualize it first, it's a lot easier to wrap your head around the concept and then you have the opportunity to drill down a little bit further. So when you're sharing things, especially if it's meant to be understood by different audiences or specific audiences, a picture, as they say, is a thousand words in many cases. And the time it takes to read those thousand words is time people don't have, especially in the business context when the things are fast moving. So anytime you have an opportunity to visualize something, you should think about how do I visualize this and how do I make these complex insights as simple as possible. Not throw everything once at a person, but make sure that they have the opportunity to be able to navigate once they're interested in more. So you really need to think from the top-down level, but give them an opportunity to have an interest. So that interest and natural curiosity will give them the desire to drill down. You also have to make sure that there's an opportunity to drill down to see that more detailed text and whatnot when it's relevant. That's kind of how you have to think of data visualization you want to get something out there as simple as quickly as possible to make it as accessible to the audience that you're working with and not try to do everything at once. So that's another issue with data visualization is like how many questions are you trying to answer with the dashboard? The dashboard should answer one major question and then utilizing actions and additional charting to get further in detail as needed to follow up on that particular question. So you have to think, is this overwhelming to somebody? And you have to think of your audience as well. If you're building to an executive versus an analyst, that's a very different audience. An analyst can spend a little bit more time trying to figure it out if you give it to an executive and it takes them five minutes to figure it out, they're never going to look at it again. Those are all these things you have to think about, but ultimately you want to make it as simple as possible for your audience to get value. And that's how you build.
SPEAKER_01I totally agree. And I think what you hit on at the end about different audiences is really, really critical. Because so we do a ton in like commercial field data, and there's not one type of person that looks at it. It's like there's a rep, there's the district manager, there's B I N A, and there's executives, right? So let's say that's four different people. Their day-to-day looks very different. A rep might need to access something by voice as they are driving around from doctor to doctor. And obviously, that's not the time to serve them up visual charts, right? That's where they need a text or they might need something in SMS, right? So that's quick hits. Or if they want a chart, they're not actually spending all day in the dashboards the same way that the BINA person is. So they need to be able to ask a question and get the chart. Whereas, like an analyst probably knows this system like the back of their hand, they don't need that. They can just go to exactly where they want, filter out exactly how they need to. So for them, it's a bit more like, okay, well, are we providing enough data density, right? Like, are we helping them answer their questions in the first clicks? The exec is like, okay, well, I want this, but I don't necessarily want this in the dashboard. I'm probably going to have to deliver a PowerPoint. If I click export, what am I getting? Does it look good? Does it have key takeaways automatically assigned to it? Right. So there's, I think understanding the user base and what their lives look like helps answer how these things can should be served up. And trying to do a one size fits all approach is just it's like an exercise in futility almost.
SPEAKER_00And for your audience, I have three GPTs I built that covers this concept. First of all, is VizCritique Pro. You put in a screen print of a dashboard or data visualization, it tells you what your intended audience is based on it, and it gives you feedback and accessibility and how to make it better and gives you actual feedback and how to improve it for the audience that you're intending without even adding additional context. You could add additional context so it could look at different things when it's scoring, but all you have to do is do an image of your dashboard or visualization, put it in there, and it automatically determines that. And because of capabilities now, Model 5.2 and ChatGPT, you ever heard of those choose your own adventure books? Of course. So um I have uh Data Quest, which is Choose Your Own Data Adventure, and then Tableau Quest, which is uh catered to Tableau. So Tableau Quest kind of goes over data visualization. So it's applicable not just a Tableau, but it would be applicable to any data visualization project. And data is more generalized. Um, both are a bit gamified. So um, once you finish a quest, it gives you a prompt for pixel art. And then that pixel, you could have it generate that image of pixel art. It doesn't do it automatically because not everybody wants to wait for an image to generate, but it has that opportunity based on your quest and how you did on data quest. You could earn specific badges based on what you're showing, and same with Tableau Quest. So it has like based on where not just one answer, but a series of answers. How are you doing in this particular area? And then you can automatically earn a badge for that. And then if you do really well on Tableau Quest, uh Tableau has something that they call sparkle. So there's a like the sparkle, it looks kind of like cross and stars for Tableau. So it's called the Tableau Sparkle. So you have more sparkle, the better you do on your Tableau quest. There's identified learning paths for both two, so you can go through an entire learning path as you would with e-learning, but in chat GPT form.
SPEAKER_01Wow. So we'll put the links for those in the description, but I've never downloaded a GPT. So where do people actually go to access these?
SPEAKER_00So uh Chat GPT, you could uh go ahead and look it for them. Also on my blog, my most recent articles have the featured GPTs, and they're all linked from any blog articles that I write. So, for example, my most recent one was Systems First Thinking, and it has all the GPTs that are feature there. In addition to that, I have Tableau Quest blog, which has a whole suite of data visualization GPTs and their descriptions as well. And of course, I'll share them with you.
SPEAKER_01That's great. I'm so excited to play around with that. And I feel like I'm I'm leaving this inspired to just try something. I have one in mind. So it's funny, my wife is actually in the process of setting up a uh like a yarn and fiber shop here where I live in in Roadwood Beach. All the expense trackers are either way too expensive or way overkill, or they don't quite do exactly what she needs. So I'm like, I bet I could like do something here. So that's gonna be my attempt is some type of a little expense tracker GPT or or app.
SPEAKER_00So what you should use for that is a GPT I just built. So it's called GPT Architect Pro. So it's a GPT builder, an evaluator, and a system. So it evaluates itself to improve its own performance based on the newest models. Because I've worked with GPT builders and evaluators on Chat GPT, but the problem with them was always it's not built for 5.2, the newest model. And also 5.2, you don't have to separate them. So when you have a builder and you have an evaluator, you could have an entire system working with itself. So you have the power of multiple GPTs and one GPT without having to switch the screen. So it really gives a lot of power to what you're doing and makes it a lot easier and more efficient to build really good GPTs. And I put about 90% of my knowledge in that GPT as far as building GPTs, there's some secret sauce because there always should be a human in a loop, and I leave that open. So people should define what they need as opposed to having something share and assume something and having it completely wrong. So I make sure that any GPT I built is there to include the human in a loop. You need to provide feedback because, first of all, when we're working with GPTs, we shouldn't be so dependent on technology, it's not all the way there yet. Two, if you do that, you're not gaining anything outside of gaining whatever the technology shares with you. You're not learning anything in addition. In fact, you may be getting a little bit slower and kind of losing what you've already learned. But this is to keep you engaged not only with the technology, but to keep you in control of the technology where you work with as a partner, as opposed to absorbing it.
SPEAKER_01That's so great. All right. So that's a natural transition to the my final question here to close us out. Um, but if you were advising an analytics or a data professional in pharma, someone who feels uncertain about AI's impact on their role, what's the one thing you'd encourage them to focus on right now?
SPEAKER_00Excellent question. So the first thing I would have them focus on is how are you going to utilize that as opposed to fight it? The longer you fight it, the longer you're behind in what you need to do as an analyst or a person that's looking to evolve with the tool, with the technology. The time of being a specialist and doing one thing the entire time is kind of going away. There's going to be a lot more people being able to do what you do because it's going to lower the threshold for those people to get in. More people are going to have the opportunity to do stuff with coding, with developing, and so forth that they wouldn't have had the opportunity to do before because a lot of the work could be helped out by AI technology. So don't be afraid of it. Also think about what you are interested in. One thing I also learned is that when you're just focused on one thing and you're doing that consistently, you become mentally stagnant. There's no inspiration. It just time to make the donuts day after day. That's one thing that makes it difficult for you to have any sort of inspiration or growth. So, right now is the time to think about what is inspiring me? What other skill set do I want to leverage professionally? And how can I extend what I use right now into that and do it organically? What should I learn on side? Should I learn how to code? Should I learn how to be a product or project manager? Uh, should I be focused on people management, building teams? Uh, should I be learning other BI tools to further enable my capabilities with data visualization if you're focused on that? Or should I be uh leveraging AI as product assistance or building with those tools to even amplify what I'm building in data visualization?
SPEAKER_01That is fantastic. So, Adam, we'll leave you leave you there. Hopefully, you also have a delicious Thai dinner uh tonight. And I really, really appreciate your time. Thank you so much. It was great talking to you, Japtain. It was a pleasure. And that's a wrap on today's episode of Pharma Sessions with me, Jonathan Kaske. If you enjoyed today's conversation, don't forget to hit follow or subscribe and share it with someone else in the pharma world who might need to hear it. For more on pharma trends, career growth, and business strategies, connect with me, Jonathan Kaske, on LinkedIn. Until next time, thanks for listening.