As we get started, I want you to think about AI not as a tool. not as a human problem to solve, but as a capability. How many of you are from out of town? Most of you. How many of you used GPS to get here? Okay. For those of you that use GPS, why did you use GPS to get here? Just call it out. Very easy. Any other reasons? Trust it. Why else? Reliable. Lack of knowledge, or it's got the knowledge I need, right? Any others? How many of you use it because it's going to save you time, right? If there's construction ahead, if there's an ambulance coming, if there's some other problem, right? Time is the one element that I've used throughout this entire experience in the last year of rolling out Co-Pilot at Pella. It's the one resource we can't get back. I can do an awful lot of things. I've been at Pella almost 34 years. We have a lot of resources available, but the one thing I've never been able to figure out how to do is truly save time and shrink it dramatically. We've had all kinds of continuous improvement activities. We'll all go out to an assembly line and save a few minutes of cycle time, or we'll save a process by going and hitting it with Kaizen events. and taking that time and shrinking it and taking out the non-value add time, right? Taking out the waste. But with AI, you can go even faster. I believe it's not just a tool you plug into your system, but it's that capability that you learn to build. It's asking the question of, how else could I do this? What else could I do differently with this to change the way we deliver value to the business? And then when we use that capability, building muscle strength, it's compressing this time across decision cycles. It's from the time I've got I've identified a problem until I can make a decision to solve it. Or from the time I have an idea and I want to go to market with it. Or it's problem solution challenges, right? From the time I've identified an issue till the actual root cause of that issue was identified. And using AI in that entire decision cycle. Doing that will make a big difference to your customers. I build this on a foundation of data and AI literacy. And it started with a program I joined with the Data Lodge a couple of years ago. It was two years ago in May that I actually started a program with the Data Lodge. And it was fantastic. It taught me about data literacy. And there was an argument back then, and it's hard to believe that was just two years ago. There was an argument of, is it data literacy or data and AI literacy? or is AI literacy separate? My answer was, yes, you are now your own data. That's the craziest thing I've ever witnessed, where you create your own data every time you talk to it. Every time you communicate with an AI, you're giving it context. If you're not thinking about the context you're applying to that AI and the data you give it, there's no reason it should ever hallucinate for you. Except it still does. Well, why is that? Well, because the models were trained on a data set that might be outdated now, even though it's only six weeks old, right? It doesn't know the context of where we're at right now. So be mindful of those things. You've got to understand how it works so you can change the way people think, engage, and act with that AI. A little bit about myself. I'm a senior data analytics strategist. What that doesn't say is, Kurt's the co-pilot guy, right? I'm the guy that says, I have a problem. Is there data to solve it? How can I get you in front of software development, senior developers of AI, AI engineers, architects, my data science team? I wanted to turn those problems that people were faced with every day into a solution. I just wanted to solve problems. That was it. And I found AI to be a really neat way to do that in mass. On our data team, we've got a data ops team that is enabling our data with AI. We're thinking of how do we make the data more AI ready? using graph, using Databricks, using other tools. And then we've got an AI and advanced analytics team that JC Hoyer, the keynote earlier, he's my direct manager. So JC and I constantly are challenged with enterprise level versus grassroots. How do I upskill the organization at the same time he's trying to get them to buy off on Okay, we want to go do a major initiative. This is going to take time and money. How do you do that? Look forward. There we go. For those of you that don't know Pella, we celebrated 100 years in business last year. We've got a little over 10,000 team members across this entire North America system. And we're an industry leader in residential windows and doors. We've tried commercial. Commercial's not so much fun. You know, they've got very custom things for every architect. every market. So residential is where we live. We love it. It's been a fantastic part of our journey through Pella. I like Pella windows because they're awesome to look through. I often tell people we compete with walls because like I'm standing in this room, I would love to have Pella windows in here, right? The view out here, the view on campus, Right? That's what we do. We compete with that process. So why AI? Why now? It really goes down to how do you lead with that AI literacy foundation and building it with people first. I believe that the people at our business make the biggest difference in our business. It's why people come back all the time to buy products from us. It's why people are impressed with the community. If you haven't had plans for this epic, coming weekend, head down to Pella. It's tulip time. It's a great time to go get some food, enjoy some Dutch costumes, have some fun. It aligns with our values. How do I align this? Well, it's because training reflects our values of being curious and being role-based. And then our AI first organization structure of how do I do this, aligning it with the vision, Well, if I don't teach the people how to use AI and use the right data to answer the right questions, I'm not really helping anybody. So I jumped in front of this last year saying, how can I do this differently? So we had a pilot program at Pella. How many of you have been through a pilot program that just ended in pilot purgatory? Okay. We didn't have that, fortunately, but we did have a challenge with the pilot. We launched the pilot and it grew bigger than the normal size. It should have been 30 to 40 people, then it got to 50, then it got to 70. I was in the pilot group and I'm like, when are we getting licenses? Because there's value in this tool and this technology. We scaled to 1000 licenses. And everything changed. That was May 1st. It was a year ago this week. And I look back on that year and I'm like, wow, what a difference a year has made. I jumped in front of it and said, I've learned how to use Copilot to save me a couple of hours every week. Let me just teach department by department, group by group, just like I would be doing in this room. Let me teach you as a small group, but like individuals. So if you've got this understanding of how to use AI, I can get you up to that, what that tech is, but it's really not a technology you need to know about. You need to know how to use it, how to build it with context, how to understand what it's capable and what it's not. And then leading with that AI literacy background. Again, passionate about solving problems, not necessarily because I wanted to teach people about AI, but I believed if I taught 1000 people how to use AI, man, what a world it would be in the following year. And it has been. So what is data and AI literacy? This is our definition. It's the ability to read, write, and communicate with data and AI in context, in both work and life. It's one of the most unique things about the data and AI literacy program that I went through with the Data Lodge. It was around not just a work skill. How many of you use Excel at home? Right? A few of you, of course, because we're data nerds, right? But when you really get down to what other things in the world do you do that you learned in your personal life or that can apply to work or vice versa as quickly, the only thing I can think of is the internet. Right? Using those kind of technologies changed the way we do everything. AI is the same thing. Building the right mindset, getting the right language. When I first saw AI, and this is crazy, 2 1/2 years ago, the big talk was on prompt engineering. How many of you remember all the prompt engineering classes that everybody would need to take? How many of you just talk to your AI now and just do it in natural language? Right? Prompt engineering was the thing. You had to figure out how to do it in so many words or less. And then the skills. When does it apply and when does it not? The challenge with most things is you want to upskill into that space, and yet not everybody knows how. They don't feel safe doing it. Selena did a great job of talking about that. That's the risk, is the human side of the emotional attachment to the way I've always done it. So I'm going to offer you this idea. This is the slide that I believe changed the way in which we deliver AI literacy at Pella. And it's going to be so foundational. You're going to be like almost disappointed in me. I start with the question of how can I do this? And I evolve it to just expand to how can AI help me do this? The key being help me, not do it for me, not do it instead of me, not replace me, but help me. That simple thing, as silly as it sounds, on a post-it, changed the way we rolled out AI at Pella. Now you might say, well, Kurt, come on, it's not that easy. For me, this is all it took. Honest to God's truth. Once I put this on my laptop and I said, okay, AI, help me write a better email. Clear my calendar, tell me what I should attend or what I shouldn't. Help me communicate up to the C-suite, down to the management levels. How do I do that? and do it for me? No, do it with me. Help me do it more effectively. Okay, now I've sent an email to my C-suite. I've got the CFO in mind. I've sent a note to my CFO. How's he likely to receive it? What's his answer likely to be? What should I expect? If I were him, how would I respond? Act like a CFO. Have a bad day. Tell me what I'm thinking, right? Like all of that became part of that journey. And by doing that, it also made me a better communicator, more willing to expose the vulnerabilities in the process, because at the end of the day, we're all human vulnerable to this experience. And the other thing I did was I taught people this. This is the first prompt I teach everyone. And I get a kick out of it because it follows the goal, context, expectations, and source that Copilot teaches. And it's a very similar prompt to what Copilot's training will teach you if you go through one of their online courses. It's write a thank you. That's the one thing. That's the goal. to Kurt in this presentation in a tone that's friendly but sounds like a pirate, which is kind of funny and quirky, right? And then use this meeting that I did for the sales organization. That was my prompt and it outputs in the pirate style, you know, Kurt, you gave a fine talk at AI Impala, right? And it walks away. with that pirate speak. Now, that's interesting, right? But why is that important? I think appreciation goes so far when you start talking to people. I like to go get birthday cards from my family, but I will admit I don't write them all by hand. I will go through the aisle and pick two or three that I like. So I'll tell people, have Copilot write three to four different thank yous in different styles so you can pick the one you like. And then send thanks. Change it, edit it, it's okay. Learn how to modify it to meet your needs and the way that you talk. And I teach people how to tailor it, to listen to them, and learn how to speak like them so that when it writes it, they're not editing very much. Most of the time they get up to about 98 or 99% accuracy, and they rarely change words. It now sounds like them. The other thing I did was I showed them how you can make a graphic and make it fun. Not everything at work has to be hard. So you can have a creative component to it. And for people that were in the marketing department, they saw this and they were like, oh my gosh, we can now use these tools to build little icons. One of my plant managers took and said, okay, we've got a safety initiative. He used it to create icons that made it memorable for safety and put those on the walls in his plant. And I thought it was an awesome use case because now he's got a tool that he can say, I want to display this. How do I do that? And I created, you know, an icon in that tool. And I don't have to get, you know, marketing involved or I don't have to get other people involved for a simple process. Another example I get, and I love this one, this is a longer prompt, but same concept. But look at the way this prompt is done. If you haven't thought about your prompting skills, the pieces I highlighted are the key parts. But the biggest challenge most people have is, I don't know what to say to an AI. I wouldn't even know how to use it in my job. You know what? The easiest answer is let AI interview you and ask you as an expert in deploying AI. Ask me three to five questions, not the end of the world, one question at a time. Slow down and answer it like it's a one-on-one conversation with your chat bot. And then Make 2 obvious and two non-obvious recommendations. The obvious ones you'll know. You'll be like, oh yeah, of course I could use that to save time when I'm writing emails or organizing my week. But the non-obvious ones will be like, and you'd be more effective if you said it clearly the first time. Right? You didn't have to have follow-ups to that meeting. Maybe you should set a better agenda. Oh, well, yeah, duh. Right? Those non-obvious ones will come back and haunt you at times. You'll be like, well, Of course, I should know how to do this. Doing this prompt pushes people beyond the scope of, gosh, Kurt, I don't know how I'd use it, but I'd love to talk with you. I forced them down this path now. You will use this first and then come back to me after you've done that. None of them have ever come back. They've all used this prompt and you're like, did they use it? Yeah, absolutely they used it. because they found that that interaction changed the way that they thought about working. What's key about this is the AI literacy component. It's communicating with that in context. If your chat doesn't have enough context, if your AI does not have enough context, it will not do the work the way you expect it to. If you get a lot of hallucinations or you're disappointed in the results, make sure it's not because you couldn't answer a next question and get it more context. How many of you have heard of the change management principle of ADKAR? I was blessed to be a part of something that Data Lodge did and they had a session and somebody came in and talked about this and I'm like, this is perfect timing. I want to roll out AI and data literacy training. How can I do this? Awareness. How do I start dripping on an awareness? Not like, hey, it's coming May 1st, but How can I start a monthly newsletter? How can I start a weekly drip of knowledge? I created an awareness campaign. So it's really as simple as, in six months, we're going to continue to drip things on you. You're going to want to come because you're going to hear it every week, and you won't even realize that you want to know more. Desire, fear of missing out. As we started to roll Copilot out, we did it department by department. I asked Copilot, I only have a couple hours a week. Help me structure this program so that I can do this because this is not my full-time job. How do I deploy Copilot across the enterprise? This company deserves to have this rolled out. We've got 1000 licenses. Let's get started. Here's the plan I've got. Make it more effective. Help me get in front of those people and set the communication standard. I would use Copilot to help set the emails out to those directors, get their teams together, and I'd start setting up one or two trainings every week. And it was a great way to start building that and building the awareness and the desire. Then the knowledge was the workshop, right? Getting A hands-on session, doing this kind of work. Ability, again, more hands-on, more Friday learning sessions. We did an every other week, Friday learning session. It was a great way to engage the community. And then reinforcement through those demos. To do this most effectively though, what I found was awesome is... I know certain parts of the business extremely well. I grew up in operations. I lived in supply chain. I know a lot more about IT than I probably ever needed to. And yet, I don't know anything about marketing, sales, or maybe even legal. I know enough to know that I don't know, but I don't know what I need to know. Does that make sense? Copilot, I'm going to be giving a conversation to Copilot on this topic with our legal team. Could you help me build demos that are appropriate for them and prompts that are appropriate for them? The minute I did that, my legal team was like, heck yeah, I'm in. I want to learn how I can use this because all of a sudden it was speaking their language. even though I didn't do it to manipulate them, but I really did do it so that I could talk to them in the language they're used to hearing. It didn't just make generic prompts then. It made it specific to them. Targeted training sessions, getting that role-based training at the right time for those users, and then getting those leaders involved. If those directors and above were involved, those teams were more effective at deploying it. Getting those users to be proactive was hardest, but we still did it by answering questions. I had this community of practice, I never dismissed a question or a frustration. How many of you think Copilot's the best AI you've ever seen? Good, we're on the same page. But who thinks it's good in some ways, right? One of the things it does really great is it covers the entire Microsoft ecosystem. I love that. But it's not the best at any one thing, but it's good at a lot of things. How do you measure things for productivity? Back to time. If I can take the amount of time I spend on things and cut it in half, we had a supply chain example where it was every week they had an hour meeting with their supplier and another hour meeting with the manufacturing team. They cut it to half an hour of each, then they cut to 1/2 an hour every other week. because they were getting more done in less time using these tools. Three C's are the behaviors that we really enabled. And I put this all into Co-Pilot and I said, help me out. I want to get people to be more curious. Curiosity killed the cat, but it doesn't kill the human. People need to know how to ask the question. Even if it's, I don't know what questions to ask. I taught people how in a live meeting in Co-Pilot being recorded, I should probably ask a question here. Can you give me 3 questions I could consider asking? Teaching people how to be curious is not native, but that's important. Being courageous is the next step. Okay, now you've thought about the question, you've asked Copilot to help you develop those questions. Are you willing to actually ask it? That takes a lot of faith that the people in the room are going to listen when you ask it and the courage to stand up and say, I'm willing to try. For me, it was learn how to stand in front of an audience and don't be afraid of it. You may not say it right. That's okay. They need to hear what you're sharing. So just keep going. Just keep going. It'll get better. And then collaborate. How do we scale this at speed? and across the business. Again, we didn't do this at Pella as a requirement. Copilot was not something that was like, you will use AI by the end of. We didn't do that at all. So we started asking, well, Kurt, how did you get 97% adoption? And that's on a four-week rolling scale. So it's never been below 97% in the last six months. Last week, I pulled the deck and it was 97% adoption in a four-week window, 84% adoption in a one-week window, and 34% of those people are such high users, they're considered power users by Microsoft's own metrics, which means they're using AI. at least 15 times each week in more than one application. Well, that's huge. That doesn't exist when somebody hasn't been told they have to, right? They got to. The community of practice was an interesting one. I was told nobody will ever come. If they do, well, it will dwindle at some point. But if you want to try it, go ahead. Great, I'm all in. I'm going to sign us up. We're going to start inviting everybody, all 1000 users every time. Do some people feel like it's spam? Maybe. Not my problem if they do, right? I record them, I save them, we have a demo, we have intermediate and basic skills. We're not getting into advanced skills at this point. But we share best practices every time. It's a place where we have two demos every time. Somebody brings a demo of, I just use it to save me time in this meeting. I use it to set up an agenda for that meeting. Or we'll get into, okay, I built an agent, and here's what my agent knows how to do. And it talks between legal and marketing, and it does this work. And we named it Mallory. right? Like you name your agent so that you give it a persona. That's fine. And then we troubleshoot challenges. If somebody says, hey, this isn't working well, well, that's a great point. Let's talk about that. If we can solve it during this meeting, we will. But otherwise, if it's more than 5 minutes, usually we are like, okay, we'll set up a follow-up and we'll share with the update. Three key ways that I delivered this was through workshops of skill development, so going department by department. I started doing internal webinars. Hey, co-pilot, help me do this more effectively. There still seems to be a gap in the way people are assuming that they understand this information. People would reach out and be like, can't it create a video for me? Or can it do some audio stuff? And I'm like, Yeah, sure. I'll do a webinar for that. And invite people to attend, and if they come, that's great. If not, it's recorded and it's published on our SharePoint site. and then blog posts. Two blog posts. One started as mindset. So mindset Mondays, I call it. And it's every Monday, I drip just a simple one or two sentences of shifting your mindset around data, data to make decisions, courage to use the right data to make decisions, courage to acknowledge if you have bad data, how to fix it. And then the second one was from another. company that I heard from and I'm like, I love the idea of a Tuesday tidbit that's more tactical. So what can I do? So that one includes all the Copilot prompts. Drop this prompt into Copilot. Let it help you answer the question. And this year was kind of fun. Copilot, how can I make this even more effective? Well, you could tie your mindset on Monday to an action that you can take on Tuesday. Oh, that's great. Can you write all those posts for the next quarter and make sure you include courage and curiosity and collaboration and drip that throughout? Yeah. I don't have time to write all these posts, right? And yet one of the funny things about it was, why does Kurt take the time to write all these posts? Silly thing is I wrote the quarters were the posts in an hour, reviewed them, agreed that, yeah, that's about the right cadence. And then there's times I'll even stop in the middle of the quarter and say, is this still landing well? Are there still people giving it thumbs up? If they're not, I'll change the last half of the quarter. But this isn't a lot of work, right? This is like I spent an hour, I developed weeks of content. And it's not about learning Copilot. This technique, which has really been interesting, is as we've developed the need for more advanced AI, and I'm sure some of you use more than one. We use Claude now as well. We've got people using Claude. We've got GitHub Copilot. But it's building that capability. The minute that they had these other AI tools, they were jumping in. and able to run successfully with it. So the techniques that we're teaching aren't really about co-pilot. The co-pilot just happened to be the mechanism we were using at the time. So how do you do this and do it well, especially when it's not your full-time gig? For me, it was iterative. Every training I gave every week, I had recorded. And I asked Co-Pilot after every training, how could I make this training better for next time? I ran a little tight on time because they asked a lot of questions. Should I pivot the context so that they ask fewer questions, or do I want that interaction? What's the right approach? One of my favorites is, how could I be a better presenter? When I'm presenting to a new audience, the hardest thing for me to do is slow down. I tend to go fast. I have a lot of ideas and therefore a lot of words that need to try to come out. And it said, the first thing you do when you get the audience to engage It's so much better if you can slow down and actually raise your hand. Do you all feel this way? Yes, good. Okay. That helped. Giving me that feedback made it iterative in the training process. And it improved that entire user experience. The teams that got taught 2 weeks ago, I'm so blessed because I now know how to do this more effectively. But that didn't mean that the teams that got it last year didn't get a good training. But they only got a good training, I would argue. So I'm going back and hitting some of those a second time. Taking them from an individual to a team. So when we think about success, everybody wants the dollars ROI, but dollars are so hard to measure when you're doing it at an individual level. 1000 employees enabled to save a little bit of time, what's that really worth? We struggled with that. But when I came back from Microsoft back in December, they came to our facility and we got to talking and they said, I think this is good. And they're like, oh my God, you're not kidding. This is good. This is amazing because you haven't made it a requirement, but you're getting over 95% adoption. With really strong adoption, people aren't just playing with it anymore. They're using it. People won't use it month after month after month if it fails and hallucinates. They just don't. They stop using it that way. But they've learned where they can use it. It's integrated now into some of the workflows. My favorite examples of workflows, an engineer took a product that took him from end to end, three weeks to do, and he does it in 30 minutes now. And it took him two afternoons of playing around with Copilot to learn how to do it. It's human, AI in the loop, human, AI in the loop, human. Three steps that he has to use and includes Excel. code written in Python that he'd never written before, and a few macros that it was enabling as well. Blew me away. He's like, I didn't know how to do this. And I'm like, well, let's make sure we do it the right way. Let me take your code and give it to a data scientist, make sure it's built right. Yep, it was. Everything's good. So at least we trust the system. You might ask, well, why didn't you just automate it? Because that wasn't the biggest thing to save that day. Automating all of those steps will come probably in six months anyway. This technology is going to do it all at some point. But right now it was, he took a problem, made a standard solution. And not only did he make a standard solution, co-pilot, when I'm done with this, make a reusable work instruction that I can teach my team how to do the same thing, and he's now taught six or seven different engineers to do the same thing. That's the win, right? That's collaboration. That's sharing that knowledge with others. So how does it work in our scale of things? It's A productivity driver. It's really measuring time. I don't measure things in dollars because people don't believe the dollars. When I tell them that last year we enabled $5 million in productivity, and this year I'm on target to do 9, they look at it and like, yeah, but it's individual productivity. So that's still, I didn't have to hire teams to do things, right? That's worth something. Streamlining processes and automation is so powerful. Finding those right pieces, fitting AI in the right spot, making the AI in the loop, and then empowering employees. What was cool about the way I taught people, how many of you like to be told how to do things? Yeah, right? None of us. I don't like to be told how to do things either. But if I can teach you that your problem's worth solving, and I can teach you how to talk to an AI, and that AI can be a thought partner. It can give that challenge back to you. It can beat up a presentation you're about to give. If it can do that and empower you to do it differently, do it more effectively. That's a huge win. And so people would listen and they would adapt and they would adopt. So what did I learn through all this? Having done this now for 12 months? Two things I would do if I could do it all over again. One is get that leadership adoption earlier. It was six months before I heard from our C-suite, hey, what are we doing with AI literacy? I should have had them on board. That was a mistake on my part, but it was more than that. It was, they were part of the initial adoption campaign. They didn't buy in. Now I will admit, I think we started with co-pilot too early. We were probably six months too early. And so they got to experience really bad co-pilot before they got to experience it getting better. And the other one is engage those AI influencers or champions earlier. We had a few people we thought were champions. They were just power users. There is a distinct difference between a person that can influence others and someone that's a power user in the platform. Don't mix those up and say, oh, because they use it 100 times a day, they must be a really good influencer. They're not always. To scale, we went, like I mentioned, team by team, it was focused on small groups, individual productivity. This year, I'm going back and hitting them as a team. How is it changing the way you work as a team? Are you sharing your props with each other? Have you created an agent to do some of the hard lifting things that need to be standardized? What about enterprise enablement? Where are we enabling the enterprise to go better and faster? That's where we're getting into multi-agentic solutions that we're implementing in places. And it's changing the way we deliver value to ourselves as well as to our customers. How many of you have heard of the Create Framework? This was a fun one for me. Couple, only a couple. Okay, this was brought to me in January. I did this same presentation to the Des Moines Data and Analytics Group, and it was kind of in preparation for knowing I was doing this talk at CIRAS. Somebody came up to me afterwards and said, hey, have you heard of Dave Burst's Create Framework? I'm like, no, but I'd love to know more. He said, we started using this to build agents. That's a neat idea. Guess what I did? I took this back to Pella the next week. I showed it to one of my AI influencers. He's like, that's a great idea. He created an agent that does this. You want to talk about power of people, right? He took a problem that I said, we need to probably have an agent that can help us create better instructions for better agents. And he built an agent that did it using this framework with the same restrictions that Copilot has of 8,000 characters. But then we also added pieces of it like suggested prompts. Build it with AI robust processes inside of it. Make sure it's got high quality. Make sure it's asking questions one at a time for more context if it gets lost. That's a really cool tool. But how many of you have thought about using AI to create more AI? The first thing we did was, okay, let's see how this works. We did a demo to our consumers, right, our community of practice. Within 2 weeks, we had 100 more agents created. We've got over 800 agents currently across the system at Pella, and most of those agents have been built in the last six months. People are building their own agents to solve their own problems. It is the coolest thing I've ever seen. They're solving their challenges with AI, not because I told them how, because I gave them the guidelines and the guardrails around it. Three sources that I would recommend you consider looking into, and I would have included others, so Data Society that's in the room. The Data Lodge really gave me the foundation to know how to do this differently. To think in terms of, I have a problem, do I have data to solve that problem? And then how do I deliver that back to the organization to change the way we work? behave. It's all about behavior. If I can change the way we behave, I'll change the outcomes we produce. But you can't get there without changing something. IIA was a partner in this space because I would say, I think we're leading in a lot of these things. Find me other people that are also leading, because when you're on the tip of the spear, you need those other people with you on the tip of the spear. It's really hard to find companies that are like, all into AI and still trying to do it safely and smartly. There's companies that are doing it stupidly, right? We see those in the news and those ones scare me, right? It's they enabled the AI and all of a sudden it deleted their database or they did something crazy in their process. And then the AI Driven Leader podcast. How many of you have heard? Jeff Woods, okay, a few of you. The idea of the Post-it note came from one of his podcasts and building an AI tool that helps me think clearly, to think about problems differently. I built my own agentic AI solution. I call it my North Star. It consists of about 15 individuals, both present and past. that change the way that I consume and analyze information. Now, does that take me out of the loop? Absolutely not. But it gives me the opportunity to hear from Steve Jobs, Michael Jordan, Leonardo da Vinci, and others to say, how can you do this more effectively? Who are you missing when you're communicating? Because you're really passionate, but are you losing people on the journey? And it gives me that confidence to go back and say, okay, let's double check that this message lands. And one thing it picked up is it added the literacy comments earlier in the slide deck. Because if it wasn't for AI literacy, none of this would be true today at Pella. It's not because I forced behavior at Pella, and it's not because I took away licenses from people that weren't using it. That's the other thing that's crazy. When you see Microsoft benchmarks, you're like, well, how did they do that? Well, they started taking licenses away. If you're not using it within 30 days, it's gone, right? We're not paying for something you're not using. I got 984 people that used it in the last week. That's amazing. But why did the other 30 or 40 not? I don't know, right? Some of them might be on vacation. Some of them just might not use it yet. That's okay. We'll take them away eventually. We're getting close. When we run out of licenses, we'll start taking away. And God be with the people that thought that they needed it later. To lead this adoption, I think there's a few key things that are most important. One is passionate for people. Make it about the people first, not about the technology. Just like we saw in the internet days, right? Like which search engine is the best? How many of you liked Ask Jeeves? Anybody old enough to remember Ask Jeeves? I loved Ask Jeeves. Why? Because I can do natural language questions. How many of us love Google? Well, because it's right most of the time, right? It gets it right most of the time. How many of them think that Google's Gemini sucks? Okay, a little harsh, right? Because Google's Gemini was coming late to the party, right, with AI. Think about high-value business cases, though. It can't just be grassroots. It has to come from the top down with high-level organizational plans as well. Big AI use cases generate a lot of momentum. Vijay's in the back. He's been involved in some of our biggest initiatives with AI in vision systems. Can Copilot do vision? Nah. right? Can it in the future? Maybe. But partnering with a company that can do that and letting us learn how to do those technologies and enable that in a way that the business hears AI more often makes it part of the conversation. And then make sure you're thinking change management to empower employees. Don't be afraid of giving your people the power to make decisions. The hardest thing for most directors in this space has been letting their people make good decisions without their permission. So if you're in one of those leadership roles and you're sitting here going, okay, I want to let my people do it, but I don't know that I trust their judgment. Let them get good judgment. Teach them good judgment. That's your job. It's not to stop them from making good decisions. Your goal would be fewer of you need to make the big decision every time. Let them make some of those decisions. Let AI help them make it more effectively. And with that, my presentation is wrapping. If you'd love to learn more about Pella, I'm happy to talk with you more about it. We're always hiring good talent. It doesn't have to be in the AI space. As we're enabling all across the organization, we are really growing, right? We're trying to find ways to grow. We're continually expanding. And I'm happy to have been a part of this journey for the last 34 years. It's been quite a wild ride. So Any questions? 5 minutes for questions. Okay, I'm going to try to get to you with the mic. We have a full house today. Well, I want to get it for the transcription because AI. So when you said you rolled out the 1000 licenses, what would you say was your buy-in at the initial point of that? And what did you present to get buy-in for that? Can I tell you the truth? Buy-in was hard. Everybody looked at me like, well, you're not an official trainer. You're not in the HR training and development program. What gives you the permission to do this? Just let me do it once. Let me do it once. I think my passion will come through. People will follow. If I start setting up these meetings and people stop coming, I'll know. Crazy thing happened. Kurt did a great training. I'd like to get in on the next one. Hey, Kurt, when are you coming to sales? Hey, Kurt, when are you coming to marketing? happened because I wasn't afraid of going to each department and tailoring it specific to them. And their adoption was, they were anxious for it. The things that were crazy were the teams that were already using an AI on the side. Marketing was using AI on the side to create content so they could create it and recreate it in our own systems. But they were able to say, oh, now I don't need to keep this on the side as a tag-along AI. And I'm like, okay, we won't tell anybody that did that. Yes, please stop doing that. But the first one was, no, Kurt shouldn't do this. And I'm like, just let me do it. We've got the licenses. We're spending $25,000 a month on these licenses. That's an expense we can't continue to do if people don't use it. And if you do it wrong, my opinion was, you did this wrong and you didn't enable it with people first, you were going to totally miss the mark. And then you wouldn't solve problems the way humans solve problems, which is let you solve your problem, not me. Let me talk to Copilot with you how to solve your problem and not take Kurt's advice. Awesome. Other questions? Oh, we got a couple. Hang. How do you make sure that the agents that people are building work together and work like business continuity? When somebody leaves, what happens to the agents that they develop that? You build an agent, you can't leave. To be honest, we've had, of the people that have created agents, The craziest thing has happened. As we've enabled this, we've had less turnover, which is almost weird, but less turnover. It's the biggest concern we have right now, though, is if we've created an agent in one department and that person moves, what will happen? We've got a process for that. We're working through that. We're looking at the Agent 365 product right now to try to build that agent registry. We've also got AWS on the wings and We want to build that solution. We've got an admin team that will manage that for me. Awesome. There were other, there was another question right back here. Very quick one. The C-suite folks that were not early adapters, are they in? Some of them, yes. So fun fact, and I was going to do a demo, but I didn't want to pull it up and do it during this because I was afraid of time. I created a podcast this year. So one of the things was, how can I get even better? One of my colleagues at another company that's doing data literacy, she said, she started doing a podcast. Great. Hey, can you be my, so I created an agent for a podcast. Be my podcast manager, help me write questions. The goal is AI literacy and data literacy. I want to talk to C-suite people and I want to also get down to directors and managers. How do I do that effectively? And I'm starting to get answers. So I've talked to the CFO and the CMO in my podcast. I've got a VP of engineering involved next. I've got a director of our product and EPMO, our enterprise portfolio management team. So as we look at these things, it's getting those people, they're buying in. But are they buying in at the rate I'd like? No, no. But they have a different problem. Their challenge is different than what an AI today of development might be. So I get it. They don't see the value yet, or they're really worried about the confidential nature of it. I don't want them to drop things in there that are confidential yet. We're still telling them, no, if it's really innovative, stay away from AI. If it's really something that you wouldn't talk to Google about, don't talk to Copilot about it either. So some of that is just the behavior. One more question I think we could take. Okay. Come on right up here. And I can stay all night if you guys want, so... So what does your data governance look like? You talked about those people that you don't want dropping that information in. Well, it's co-pilot. Someone, everything that they drop in. So if they drop in a document to co-pilot, and we give advice on it, but can you really monitor it? Sure. but it's only through their own tenant. So we've got Microsoft admins that are watching for really nasty behaviors. We're building an agentic solution for that. Most of it's been all on your person. So it's education. And every two weeks when I have these conversations, there's always a security conversation. I bring the security team into that all the time. The other thing has been our master data management and data governance side. So as we think about building tools, are those AI enabled? Is it a single source of truth we can trust? But like any 100-year-old company, we've got data that we don't like and that AI really doesn't like. We're working through it. Awesome. Well, I think we're at the end of our time. Kurt, this has been extremely helpful, extremely valuable, great to hear the real world experience and how you're implementing. Are you sticking around for the rest of the day and networking sessions and things like that? So please, let's give a round of applause for Kurt. Appreciate it. Thank you.