Why Starting with the Problem Drives Better AI Tools for Rewards Teams

This is the range podcast where comp and total rewards leaders share what they're building with AI and how it's changing the work. In this episode, we're getting into what happens when a rewards leader starts with the problem, not the tool, and ends up with a custom AI agent his entire team uses every day. This one's for comp and total rewards leaders who are done waiting for IT to tell them it's okay to start. Welcome to another episode of the range podcast. Ryan is a director of total rewards and people operations at Edmonton. He's been around for 15 years across manufacturing, tech, retail, and financial services. He is known for his behavioral economics approach through Warren Design, and he's an AI builder who designs custom agents for his HR team with zero development training. In his own words, problem first, tool second, Ryan, welcome to the show. Thank you. I am thrilled, Jack, to be joining your podcast here. All the rangers going to coin that term, rangers, folks that I want to hop on for the right, and I am thrilled. I am also very caffeinated, so buckle in, my friend. It's in the morning here in the US, and I realize you're probably on the end of your day, they're across the pond, but we're going to have some fun today. Absolutely. Now diving deeper into your music taste, Ryan, if there was one song that really pumps you up, puts you in a good mood, I'm curious what would that be. This wouldn't necessarily be something I would want as my theme song, like walking on to a stage or anything, but walking on sunshine, you can't help but love that song. Like you hear it, it starts with the drums, and then like you can't not move to that song. Well, if you're not moving, you're either asleep or you're permanently asleep. I'm wondering, you got three kids, you got a wife, you got a mother-in-law that lives with you. Where do you find, not just where do you find the time, but how did you come across the eye? How did the journey start? I am self-taught. There is no one out there that is going to eat, equip you for all this, and I was really inspired a year ago. I went to the World at Work Conference here in the US, it was in Orlando, Florida, and as part of the other national conference, there was a presentation done, panel folks using AI, what's next for AI, and one of the presenters was an age of BP from a large pharmaceutical company, and they were telling us about their GPT that they made at that time that helped them with their exercise routine, they would take pictures of their food, and then their GPT would tell them how many calories, protein, and so on, and it would log it all into a spreadsheet. And I thought, well, that's kind of a gratuitous use of AI, but honestly, how hard could it be? So I came back from that conference and I'm like, okay, well, I'm going to build something, but I'm going to build something that I mean, not a gratuitous use, so the company I was at at that time, one of the challenges we had was we didn't have job descriptions everywhere. So one of the ways that we kind of meet the business where they're at is, well, I made a GPT that, in a conversational way, using words, walk leaders through how to write a job description. So I built into the backend, write a job description. It was not just enter this into the template. It was tell me about your job. One of the examples was we had a leader that had a task of 80 things. This is what it is. And I said, well, just take your list and put it into the GPT. I didn't know what would happen, frankly, but what happened surprised me is the tool took all of the items and indexed them and then using the large language found market benchmarks from each of these surveys, but then not just gave the leader the benchmarks. That was behind the scenes. What the leader saw was, here is a JD that resonates back to my listed 80 things and then here, like, I walked through so how much discretion, how much independence is there. And the tool really amplified what the leader was doing. No one wants to write job descriptions, right. So it was spending less time at the keyboard and more time thinking about what types of jobs do I need in my team and organization. That's where I got really surprised and jazzed about what these tools can do. And that's an excellent way to start. You looked at people were using AI, you thought there must be a more relevant way to use it in the tool reward space. And job descriptions are probably a really safe place where you can start experimenting, right. Because you're not necessarily dealing with any personal and identifiable data. And you're tackling a clear need, right. No one to point no one wants to write job descriptions. I mean, there are a few people that love writing job descriptions, but they're very rare. You know, we all need to understand our lane and life, our skill set. And when I start with this, I started with what's the problem I have and then came up with a tool to fix the problem, as opposed to thinking, what can I do with AI? It was like, I looked at, so what something I could solve with AI. And at the end of the day, what was really cool about this GPT, again, it was so a year ago, 2025, like it amplified the manager's domain expertise. So the manager knew the specific like e-commerce roles that they have, the specific benefits or specific manufacturing roles, they know what they want them to do. How can we help them articulate it and bring it to life through using just common natural language as opposed to helping them or doing any of you with them and then us trying to write it, like it removes the middle person, but it also directly meets folks where they're at to amplify the other domain expertise. Gotcha. And looking back, Ryan, how do you think this, I mean, you were a rewards leader a few years ago. Now you are a rewards leader that now there's also AI enabled. How does it feel like? What is it? How does it change your day to day in the good or in the bad? Like, tell us a little bit more about how it feels to be an AI enabled. Either what it means for you. We are living in an exciting time. I am thrilled that that we have the tools today that we do it. I couldn't do my current role across all the domains I'm responsible for from comp to Ben, to HRIS, to HR operations, HR compliance payroll. All that fits underneath my umbrella. I have a team, by the way. It's not just one person, but I also have my custom agent as an additional team member to help with the lift. And so I couldn't do what I do today without AI, but at the same time, it's not that I am abdicating and just trusting what the AI spits back. It is all about, you know, trust but verify. So, AI is like this bright, I wish you tailed intern. They want to do lots of things they can, but at the same time, you also need to to check the output. Make sure that it's working for you. And along the way, as we think about here and in the reward space, this is, this is, Arya is not a tool that's just mine. I also have it shared with the rest of my team so that we can all will leverage the intern together while we all might have the same code. Each one of my team members has, so, so Arya, who's Arya? She is my automated rewards innovation ally. So, AI. I'm also a big game of Thrones fan. And so like all this kind of just locked in for me, but everyone on my team uses this tool differently because we personalize it. So, what does personalize me? It's not saying, hi, I'm Ryan. I have three kids and and I help with with Mother's Day. It gets back to thinking, what are my goals? Who do I interface with? What types of projects are ongoing? And it's kind of like a little bit of a chief of staff also a little bit like an intern, also a little bit like an analyst and analyze files or move files around. It all of us use the same set of code and skills. At this point, I think we have about 20 skills built from making just a job description to market pricing, to analyzing, to moving files to creating dashboards. But what I love is that my team is also making skills too. Yes. For what we're talking about with skills, so that's where you have an enterprise from that perspective. And so, for folks that don't even know what a skill is, let me just kind of anchor us here just a little bit. So, we have within co-pilot. I think co-pilot we have agents. And co-pilot agents can do different things. You can make your own custom agent. If you have co-pilot studio, you can then even do bigger things with your agents. Then within the open AI or chat GPT area, they have GPT's. So, those are kind of like another agent that you can build a certain purpose. It can do different things. So, that's a year ago where I was really starting to learn, was with chat GPT and building a GPT. GPT that live with an open AI, but then Anthropic or Claude is another big one that's out there. And then so Claude has a number of different ways that you can work with it. But Claude also calls them agents, kind of like co-pilot. And then Gemini has another version that they, and so I'm not going to go through all of them in the interest of time. But, you know, there are different ways that you can personalize AI to the work that you're doing in rewards. But now, ask me about privacy. What about privacy Ryan? I'm so glad you asked. So, I like to use the eyeball rule, where if you can see it with your eyeballs, so can AI. So if you're asking AI to open up a file to tell you what's in it, well, at that point, like, like, everything that's in that file is accessible through the AI. But if you ask, for example, AI to move a bunch of files, can I show you a really dumb example? I'm going to share my screen if that's okay with one of the other dumb things. All right. We love some action on podcasts about AI and Tori Lawyers. All right. So this is something that I shared on stage at the World of Work conference this year, where I did live on stage, unscripted a debate. And then the audience gave me questions that give back to my agents. And also recently I went to one of a local business schools here called Carlson, a college in the US. And the professor there also asked a question from the final exam. I was a little bit nervous already passed, but I'm going to show you one of the dumb things here. So this is a dumb thing. And we need to have it all added to one folder. Okay. So we're talking like 140 plus different files. They were all saved within different folders. On the left side, you see this black screen of death. This is my blog code interface. Right. So I'll just say this. There's a higher learning curve when it comes to cloud code, but the ceiling is limitless. So within cloud, there's a lot of AI, which goes through your browser. You ask it a question that gives you questions back. Then there's cloud co-work and desktop that is similar to cloud code, but I can work with files on your machine, but as a nicer interface. But then there is, and one nice thing about cloud desktop is that you can set up projects to share with your team that each of you can work within. Then there's cloud code that you see here, um, cloud code. Again, a little bit of a steeper learning curve, but this is all the developers, all the folks in IT when you hear cloud code and you're going to take it from my cold dead hands. This is mostly what they're using. And this is where I built Arya into here. So this is my request to Arya. This is the tea up here. So for these three years, I need the payroll files. Then there's going to be, you know, I want you to, they have all the same file names, but they're in folders by years. I want you to pull them all together. Oh, and by the way, we changed HRIS systems and in 2025. So there's going to be different file names. This is what it is between the two. And so I'm just gonna, I'm not going to play the whole video here, but so on the right side, you'll see here's an example one of the folders. Here's the pre-check detail, the payroll register, but we need to bring that into a centralized folder. And so here, I'll just kind of scroll through. There's all these by all these years. And then so I ask Arya to go through. And so she starts writing a script and it explores the folders and then ran to an error. Then I then rewrote the code to find things. Okay, here's what she found initially. And she asked me, before she does anything, you won't do me go ahead and make the copy. Well, I actually have some other context for her that, hey, make sure we know which ones are off cycle versus regular. So add this in there. And don't rename all the files the same. So moving right along. Now she starts writing the actual Python script. I use Python every day. Yeah, it was about to ask you right. And like the car code, Arya caught an error in the code. So it started troubleshooting on its own. How important is it for a person to actually understand that line of code? Like, is it, is it something you pay attention at all? Or is it? So I pay attention to the output. Did what I asked my tool to do? Did it actually happen? And, but I also have it check itself along the way and audit. So as we saw here, first she went Arya went and found know where all the files were. But then she before she did anything, she came and asked me, do you want me to make any changes? So this is an example of not just human in the loop. So claw code will ask the human, are you okay? Are you okay? Or K? And multiple hit the one button or the two button just to keep going through. But in this case, this is an example where it's humans deciding. So she pulled this stuff together. This is what she found. But before anything happened, I saw the initial dry run. I validated it. And then here at the end, in just seven words here, go ahead and execute the copy. So as we see over here to the right, now this is the payoff. So I'm just going to scroll through. It's going to run the code and bam. So on the right side, you can see all of the files that were just copied over all 140 and so on like like this. This was the payoff. And ask me how long did all this take? So so even exactly. Right. So because I look, I think in many people's heads, when when we talk about using AI in comp, they're thinking like these folks are building an end to end, a new end to end compensation platform replacing all the vendors at once, all the pay surveys in a couple of minutes or maybe a few days. I mean, cool. If you want to run with that and try it, test it out. But really, dealing with here is a tedious time-consuming task that you could have otherwise done in days, if not weeks, because you've changed your HRS provider. So just think about the amount of time required not just to retrieve all the files, but then find like emulginate them together, find like a consistent template to do the analysis. Whereas are you did it in how long? How much time? Seven minutes. Seven minutes, the new record. So and but no, this was just the version I recorded. And the version in reality when we went through this, it was me, my payroll analyst, we had this request like how are we going to figure this out? It took us about 20 minutes because we, we iterate a little bit more and there were actual files missing. So as I went through and check the folders, we're missing these registers from this one or or and this cycle, we had multiple final version one, final version two, final version three, right? And so and and actuality, I took a little bit longer, but I think air quotes here, it would take in my analyst at least a day of just mindless clicking, bringing it over here, right click rename. Now we could maybe written a script if we were in IT, but again, we aren't techy. Like I don't know how to write a power shell and Python, but like this is the and the the use case of, you know, how can we spend less time at the keyboard doing my numbing things? This is an important ask we had to deliver, but we were using a tool to help us get there that also self-autical on the way and the file paths, not the contents of all the private employee information in the actual payroll registers. So no, the banking information, none of the employee name, mailing address, all that. And this example, this is an example of how we can think about using a tool like this, but also maintaining privacy, again, going back to my eyeball rule, I in this eyeball rule, all they saw with the folder paths and the file names. And so as we balance between privacy and being able to do what what what we need to do in an efficient way, one thing that I found super helpful is I have good relationships with my IT folks that I I regularly check in with friends over in info security over over in our cloud area, just so that they're aware of the kinds of stuff that we're doing because everybody is on a continuum right now of learning. And the goal here really is to bring friends and colleagues along for the ride as we look to use these tools because all of this is new even for folks in IT and think about it from a change management lesson. So I'm going to bring my behavioral psychology into here a little bit and if we, you know, think about change management, we need to bring folks along for the ride so that the change doesn't feel like it's being done to people so that it's not initiating that fight or flight response, but that we are bringing folks along for rights or amplifying what we're doing. But I think by humans are still deciding what happened. So it's not human in the loop by humans are deciding. Yeah. So were you and you said that your team, you know, it is adding on our it's adding new skills, adding features. One of the questions at for you is like, you know how one of the key questions you get from rewards folks. Okay, you're introducing your tool. But what about like who's who's going to maintain it? And what does it mean to maintain it? Does it create new workload? Does it does it actually, you know, is there is there anything there that we can say about where the implications are? We get the time savings. You've saved days on a job that, you know, you've already has done in seven minutes. But what about like who's maintaining Arya? If it's if it's everyone, what are the implications having like that the central. model to maintenance or does it maintain itself. Yeah. Yeah. So one of the things that I was toying around with early on is as I was building ours, she said, well, I could have a shared consciousness that where I could live in a space between everyone's machines. And I'm like, I don't know if I'm ready. I've seen that movie before where in theory, I could have put all of her files on her skills in a shared location and then have everybody run her locally, but then she would be thinking across all of our machines. I wasn't quite ready for that. So, but as we look at how do we maintain this right now, I've been the main maintainer where we basically share skills we've built. So every week I sit down and have a huddle with my team. Let's talk about AI. What have we done this week with AI? One, so that where I am bringing folks on my team along for the ride because not all of them are builders, not all of them are super jazz, but as we look at individual use cases, like look at all these files we just moved or in other cases, hey, I made a skill that can identify the job level for a job description. Cool, right? But as we kind of figure out how do we maintain this, I don't know yet. I'm going to be honest. I don't have all the answers. I don't think anybody truly does, but I am. And these kind of custom agents, like, didn't exist three, four months ago. And one thing I did know is I didn't want to have to share consciousness thing. That was a step too far for me. But as we have a shared location where if someone makes a new skill, we go ahead and throw it out there. But some of our skills are tailored just to us, like I have a skill that's just Ryan's voice. I am the voice of Ryan Berkey. Ryan likes to go for long walks in the beach. You're afraid of people impersonating you? If they have access to that skill. So if they can get to my hard drive and they can get to it, sure. God bless them. But am I am I afraid of them impersonating me? Not anymore than anybody else kind of kind of trying to impersonate or spoof somebody else, right? And then the day it's about being the real you, but also not advocating. Like I don't have Arya right and send emails on behalf. She can treat you as my inbox all day long and an outlook. But she doesn't write messages and send and reply to correspondence. It was it. So it kind of gets back to where do you have your own personal boundaries with tech and and how you use it? And I think you know everyone thinks the hard part is learning the tool. It's not the hard part is knowing what to build. So thinking about what is it that what's the problem I want to solve? And how can I bring a tool built something to fill the gap? And in the case of of Arya, you can even just do it with just generic cloud code. You know, but there wasn't anything big that I built. I just described what I want in English. So as we think about, you know, access to these tools, there's this term called radical agency. A radical candor, you're being candid, but but but where I am, I am kind of pipping a bit is radical agency where AI is collapse the barrier between idea and reality. You don't need a development team. You don't need to specialize knowledge. You should bring your IT friends along for the ride. But you don't have to ask for permission to try and build something whether you're building it with co pilot, you're you're making your own first agent there, or you are at and a I touch a PT shop and open AI shop. At the end of the day, it's just going out there and trying because you don't have to get permission to do it anymore. I mean, you need to make sure that that you're not that you're having good data governance in place. But but right now, we're only limited by what we can imagine. So as we think, you know, the doors open, we can walk through it. No one's here to tell us no. I mean, no one really wants to be told no at the end of the day. But we're we're limited to what we can envision, which for for a lot of us, it's it's limitless. Like if you can envision it, you can build it. But we also need to make sure that we don't get give up what makes us tick. What makes us human? We shouldn't advocate our thought process. So can I tell you a dumb thing I did with AI? This is probably the dumbest thing I've ever done. I think we all want to hear this like I'm bearing the lead. I'm I selling it enough here. So and where I'm in the US and Minnesota, I was planning a trip to go to Wisconsin delts. And this is end and December. And I had pulled up check GPT and I was using my my pro account to model out. Okay. So where's the train? When's it going to come in? Where's it coming from? Are there any weather delays? I had built matrices. I had multiple instances and dashboards coming up. And I was like, okay, we need to get to their to delts by this time. But because we want to get there for reservation. So should I Uber from here or there? And I I had spent the whole evening just trying to map all this out. And I was just sitting there waiting for the next response from AI. And my wife walks into the room and she asks, what are you doing? I'm planning the trip the trip. And she sees all my screens open. And she just sits down the couch. And two minutes later, she just shows me her phone. Here's the answer. Just buy the tickets. And I will feel I felt a little bit accused in that moment. I thought I was doing everything right that I I was maximizing. But what I was actually doing was I was advocating my thought process to AI. Yeah. You're like, you're a super user. You think you're doing everything right. But in the day, the perception. Yes. It sounds like you run into this Jack. Every day, I get a constant fight right. And honestly, and the and the context switching. I think there's going to be studies 10 years from now that show the damage we're doing to our brain. It is really important to stay like to be able to stay focused on one thing and know what you're doing without delegating too much. Because yeah, that that's a bit of the like the natural instinct we get, right? You said that imagination is the genuine requirement. You know, it's knowing learning the tools fine. Like there's always the tool can teach you the tool can teach you. But it's knowing what you want to do with the tool. I think that's the key message, right? If we if we look ahead, if you can think about like a genuine opportunity you see in the total rewards domain or even just a frustration you have with how things have been done dealt with in the past, you think there's a real opportunity now to change and to move from old ways to newer ways. What would that be? Like we're we're in other words, what do you want to build next? Where does your is your imagination taking you? So it really gets back to what's the problem you're you're solving for. And I want to just circle back a quickly about kind of advocating because every tool changes you. We used to print out map quest maps back back in the old days of the internet to get from point A to point B. Now we use the these these these magic screens in our pockets phones GPS to get us there that that tool gets us there faster. But if if if we needed to if our phones stop working, we need to get back to where we were. Could we do it? Yeah. Wouldn't it be sufficient? No. And every tool changes you. But but what's exciting is that we get to choose how it changes us. So it's not that that we have all this context kind of switching is going to make us dumber. AI is not going to make us dumber. Well, it will if if we abdicate our critical thinking and discernment to it. If you allow it to exactly just because you have a piano doesn't mean you're a musician just because you have a set of sculptor tools right the more you use those tools, you're going to get some calluses on your hands. Every tool shapes you. We don't use map quest anymore. Well, there might be somebody out there. If you do, go down here for using map quest. But as we think about, you know, what we can imagine, what I'm imagining next part of it is thinking about what are some of the things or or challenges that I have as as I have been using Ariana for the last four months. I have a number of different meetings that she has had notes for me on for over last four four months. And one thing that I'm working on is a historian skill where she basically is going to be compiling up an ongoing notebook of all the stuff I've worked on indexable so that I can search through it. And if I am getting ready for a meeting, she can put a dossier of, here's all the things associated with this and potentially that I might want to consider as as I go into this meeting. And so it's still I am I'm building. But it's it's thinking like, okay, I have all this stuff. What can I do with it? Now it's it's asking what whatever what can we as opposed to what what what what what what are we only allowed to do? Yeah, I used to I used to have a diary when I was a young kid. I used to write whatever was in my mind and learn more about me. And then I stopped writing my diary more than 20 years ago. But one of the things I use the eye for is quite similar to what you described there is to populating my story bank. So as I continue you know going through my living living life essentially I'm going to be 36 years old tomorrow. So I'm starting to have a few more stories. Thank you. Every cult transcript exists in my files. Everything I said about my myself, my career path, my story, my interests, my passions. Cloud goes in there, extracts it and adds it to my story bank. And it's quite powerful because you can do so many things, I'm not playing or writing a book, but as you know, we get all the content out on LinkedIn. And the way to make it feel yours is if you actually open up, ensure more about yourself, because any AI can write a post. It's only if you include those personal traits and stories, that's the moment when the content becomes authentically yours. So that's how I use it, but I can visit one day, maybe my agent, story bank agent, talking with your story bank agent and writing a book together or something else, like whatever day you're in. I'm sure it'll be a comedy. It'll be a comedy of errors. It's going to be like dark stand-up comedy show, I think. It's going to be more like a Catholic way, that kind of thing that you cannot delete from your memory, but at the same time you're amused, that's what I imagine, that's what I imagine. Yeah. That is, you know, as we look into the future, what I also envision is that everyone's going to have their own bespoke agent, that knows their voice, that they kind of take with them, that kind of is part of their CV, we'll see what happens. I'm saying this now in 2026, when I listen to this again, and our agents are making the comedy book to be like Ryan said, this boy was he wrong, or we'll see what happens. But I envision folks having their own custom agent that they take with them, like their CV, that is some skills, tax, and so on, like today, folks know, like if you're living in an Excel, you kind of know like you're go to formulas, or you're like this is a chart, I spent hours making, and I take a rinse and repeat this for other reward cycles. I guess this is kind of the Ryan Burkey framework as it were for communicating, paying performance. Others kind of have these today kind of an Excel templates as it were, but now it's not going to just be an Excel template, but it's going to be your own agent that's going to follow you and grow and get better as a thought partner, not someone that's going to replace you. Again, it's humans and AI together, not AI replacing humans. They're building tools for humans, so I mean, yes, no one knows what can happen in the future. You know, there are different forces in the labor market. There's not just AI, but what we know for sure is that Arya was it built to replace your analyst is actually a tool you've built for your analysts in your team. And you're, we can say like augmenting or amplify. I think you prefer to amplify them, right? Make it better. It's about, you know, increasing our productivity, but mostly it's also about, you know, going down to why our function exists, improving the employee experience in the company, the manager experience, and improving the allocation of resources in the company. So it's a win-win situation. And you still need a no-stuff, right? You still need to have gone through your own personal struggle, because the more you struggle, the more you build roadways into your brain. There's been MIT studies that have shown that kids that were just giving Google and kids that were given an AI to help them write a paper. The AI papers were polished, pristine. They were great. There was a level work. Whereas the other students that just used Google weren't as great, but what they found when they asked the students later on about the content of those papers, the kids that used Google, their brain scans lit up, whereas the kids that used the AI, their brain scans didn't light up. And so this kind of gets to the point of, you still need credentials. You still need to know and have a certain body of knowledge in order to, you know, discern the information coming off from AI. Is it AI slop, or is it work slop? I like thinking about, I like work slop, where it's not about, if you're going to take it and you're going to use it, you're not putting your name on it and it's up to you to know what's behind it. And that's where like education, certifications, credentials comes into play because you still need to have that background in order to stress test what's coming off from AI because if it looks very polished and AI is super confident, that's the time when you should super push back. And the only way that you know to push back is if you have the other credentials you've gone through the schooling, it's not that you necessarily need to have gone to Harvard. You could go to a regular university, but it used to be about hiring the smartest person in the room. It's no longer about that. It's about hiring the person that has the vision for the future, but that vision coupled with discernment to be able, because not everyone should be, I'll be hiring artists without discernment. We shouldn't hire critics without imagination either. Right. So going back to your previous, you said, this is what I got wrong with AI. What's your vision for your next vacation? And are you, are you going to promise your wife, you're not going to use AI to plan it? No, I am still going to use AI, but I'm not going to just sit there and wait for AI to make the decision. I'm using AI as an information and potential options as opposed to waiting for it to make the decision for me. That's where I found myself advocating was, all right, this is the one thing, I'm just going to let it optimize and let me know. Well, I'm still using AI, but I'm also deciding what's going to be happening. Having that watershed moment, that self-realization that I am going to make my own decision, because this kind of gets to my Catholic card moment. So I'm a nerd, I like Star Trek, and I like to use the analogy of Captain Picard on the enterprise. He has all of his officers, he's a science officer, his weapons officer, engineering, his communications. Well, you know, here's a scenario, aliens are coming, they're going to start shooting at us, but we need to react, we need to do something back. So all of the AI are telling Captain Picard, you need to do this. What does Captain Picard do? He sits back in his chair, sits his Earl Grey tea, not even coffee, just Earl Grey tea. And he, meanwhile, the AI's are all telling me, you gotta do this, you gotta do this, you gotta do this. He sits back. And what ends up is that, no, this was a computer hallucination, it was a simulation. If he had actually fired the weapons and response, he would have caused World War, galactic war, whatever it is. And so, in this case, like, we need to be, you know, we need to discern what's coming off from AI, but we need to be courageous to make the call, the models won't. And then the AI won't ever care about humans. I mean, I mean, chat GPT will definitely want to try and be your friend and all, but in the day, like the, the human connection to what the decision is and how it impacts others. AI is great about optimizing options and situations, but it can't care then the day. And we need to be courageous and have the guts to make the call that the models won't. So, like, we have access today. If you can vision it, you can do it, but we need to stay sharp while we're doing it, like, like, not give up our thought process. We need to judge what's coming out from it, but then be human. We need to have the guts to act on what we believe. As T.R. leaders, we have been in the room for five, ten, fifteen, twenty years. We have earned the place in that room, not because of how we've used AI, but by how we have shown up to leaders, we have that credibility. It's that gut feeling you get that makes us all human. It's also that gut feeling that Ken Bacard had as he was sitting with his Ken Milti. He knew better. He didn't know why by, by, he, he knew better and, and it's important for us to recognize that AI is here. It's, it's, it's, it's not going anywhere. We need to say for a while, we need to use it, but we also need to know when we just don't have to follow it blindly. Because at the end of the day, that, that's what do, do the thing the models won't predict. Sorry, SkyNut. Not sorry. That's a perfect note to close the podcast episode, Ryan. Thank you for walking us through your philosophy, how, your story and how you started. Thank you for being in the podcast and we wish that other leaders getting inspired from your story and Ken start building cool tools as well without forgetting about applying their judgment and trusting their guts. Thank you, Jack. This has been fun.