Christopher Parsons | Building the Learning Organization of the Future: Human & Agentic Capabilities

So I want to do a little talk, mostly about the future, but many of you know I'm a history major, so I want to talk backwards a little bit. So this time last year At K Connect 2024, we had just put synthesis video 2 out to everybody. So. So it's true. That was just a year ago. I looked it up. It was August 5th. That feels like a lifetime ago to me. I don't know if that's true for you all, but like, so much has happened since then. So At K Connect 2024, we had talked about the fact that we had just started developing AI search, that we were starting to design synthesis lms, and that we were just launching a learning and development client roundtable. So that was a year ago. Since then we've done a lot of things. We've got 125 people in our community in the public beta for AI search and over 65 are live to all of their employees. We have, as many of you heard today, a private beta for lms. So we've got seven firms in there and we'll be adding more folks in. We have been writing and talking our way through change. And so in the last year we launched the Smarter by Design newsletter as well as a series with Troxell that we partnered with called welcome to KM 3.0. And for me, maybe some of you seen, I've been More active on LinkedIn a little bit too. And to me, when you go through a massive disruptive change like bringing AI into knowledge management and now into learning and development, there are just so many questions that get raised. And it's really helpful for me to write my way through it, speak my way through it, have conversations with you all, bring people onto the podcast, interview them. Like that really has been a core part of what we've been investing in over the last year. I think it's also interesting to note how much community wisdom has been surfaced just in the last year around AI search. So obviously starting with the talks we've seen over the last two days, but looking at webinars that we've done, client roundtables, people sharing stuff with us one on one, like, this community has really learned how to take, you know, AI search and deploy it for knowledge management purposes. Like, we've learned a ton since K Connect last year, you might. Carrie had the slide right. AI plus KM equals fire. We didn't talk about AI at K Connect 2024 really at all. And I promise you we would talk a lot about it this year. So promises kept the other part of community wisdom I've been really enjoying is around lms. This is obviously behind what we've been doing with AI Search, but there's been a lot of insights shared this week specifically around the fact that it's not just a technology change, that's maybe 20% of it. It's really culture, it's process, it's people, it's figuring out what's our philosophy to not just an lms, but to learning and development writ large. And so this community has been really active already in sharing and I think that that's really gonna keep accelerating. So what I wanna do is kind of imagine that we're here at K Connect in 2026 and kind of talk about maybe what we'll be talking about. So I want to break that kind of into three parts. I'll talk about what we'll be talking about in terms of lms, slash learning and development, what we'll be talking about in terms of AI. And then the most fun part is all of it mashed together. So from an LMS perspective, we are releasing learning paths. As we talked about for those of you that went to the future of synthesis LMS breakout, we talked about the fact we're doing a launch in Q4 around supporting self directed learning paths in Q1, we're releasing assignments and recommendations. We've got a whole list of things that we're gonna build after that, but we haven't figured out exactly the order and the reason why that we're committed to building, but we haven't scheduled yet. Because when we're doing something this new, when we're going from zero to one building a brand new thing, we really don't wanna go more than two releases like six months out, right? We don't wanna make promises going out a year, a year and a half because we learn new things early on in a process like this, in a beta. And we wanna have the flexibility to respond to the community and build the stuff that matters most. We have made the mistake in the past of kind of roadmapping out too far and, and kind of saying we're gonna do this clever thing and then that clever thing which enables us to do this clever thing. And we've learned through experience that doesn't really work very well. One, I think we as humans aren't as good as predicting the future as we think we are. Conditions change, new technologies emerge, new facts emerge. And so what happens is if you make promises out too far and you get new information, you're left with Kind of two bad choices, bad choice. A bad is you keep your promise, that's good because you keep trust, but then you build the wrong thing because you don't want to break your promise, that's not good. Or you break your promise, which isn't great for trust, but then you build the right thing. So really our kind of point of view is we're going to be as transparent about what we're building. Kind of two releases out, six months out, and then staying very flexible to understand what the next most important thing for us to build is. So another thing that's been interesting in the community is to think, well, given such a dynamic, evolving situation that we're in, so we've got AI in, we're building a new LMS product. I think kind of culturally there's some opportunities to re examine how we do learning and development. One way to think about it is most of the ways that we do learning and development kind of in aec, but just generally they're all pre AI, right? All the ways that we think about, you know, doing learning development in an organization there before we had AI. And so it's, it seems to me that it's quite obvious that some things are gonna change. So if you can now more confidently get information in the flow of work, maybe we don't have to do so much upfront. Kind of just in case you ever run into this training, right? We can just have confidence that people can find stuff when they need it, and we can be more targeted in things that help people get broader understanding. We plant seeds, we explain philosophy, we explain values, but we don't have to necessarily teach people all the details up front. So I think that's true. However, I think there are limits to this. So I wrote about this a few weeks ago, but I was listening to a podcast recently with the CEO and founder of a legal tech company. They were building an AI product that's kind of a lot like what we've done with AI search. So for law firms, you upload your contracts, your policies, your procedures, and then people can ask questions of it. And so the CEO was being interviewed and the guy interviewing him said like, okay, I've got a real world scenario for you because this just happened. A client of ours offered me hockey tickets and can I search your product to find out if I can accept them or not? And the founder said, absolutely. That's what this tool is made for. It'll search your policy and it'll give you a citation. It'll tell you exactly where it was found. And I'm like, cool, makes sense. We're doing the same thing. And then the guy interviewer asked him, so does that mean that I don't have to sit through these boring HR videos every year that tell me all of our policies and procedures, like, can I just use AI? And the founder said, absolutely. And I disagree. Here's why I disagree. Yes, it's true that he can search in AI and find the answer to whether or not he can accept the hockey tickets and find out what the policy is. What he doesn't learn from AI is that he should ask the question in the first place. That's what's missing. That awareness that it's actually a thing around ethics to take gifts from or whatever the policy is. We have to have this awareness that this is actually a thing that should be planted in my subconscious. So I think a good training maybe in this L&D plus AI world, gives you foundational ideas, plant seeds, let you know who to go to or where to get information, but you don't have to memorize the details. So in this example, maybe it turns out that the policy is you have to like, report gifts or fill out some form if it's over $500, but if it's under $500, you don't. I'm making this up as I go. You don't have to remember in that course that it's $500. It doesn't matter. You can go find that later. It's more planting the idea. You need to ask the question in the first place. So I think AI does not. I'm sorry. Oh, AI does not mean that learning and development is dead. So that is just like a small example of the kind of questions that we're gonna be asking. And we don't know what kind of technology we're gonna be building exactly as we have this conversation as a community. But we do know what our vision and our goal is, and that's to help connect the right person to the right knowledge at the right time. The right time might be in the flow of work. The right time might be before they get started working at the company, before they get started in a new role, before they get started in a new project type, before they get started in a new phase of a new project type. That's what we think is so powerful about this new learning management system we're building, is really getting more and more knowledge to the right person at the right time, whatever that is. Okay, so that's a little bit about learning and development. Well, I will say, actually, one more thing. It's raising a lot of questions and questions that go beyond product. The poet Rilke talked a lot about living the questions. And so between now and K Connect 2026, we're going to raise a lot of questions as a community. We're not going to be able to answer all of them in the next year. I feel like this is a change that's going to take longer to roll for the community. We'll answer a lot of them or we'll get some good ideas. We'll get beginnings of good ideas. But this is going to be an active conversation as a community to figure out how to become a more modern learning organization. Because I think that's what we're really talking about. I think there's a sea change coming in learning and development, and I think it's all about modernizing the way that we think about learning, development. Does this resonate? Is this making sense to you guys? Okay, that's good, because I staked the future of the company on it. No joke. All right, so from an AI perspective, we talked about Advanced Project Search a little bit. For those of you that weren't in that Future of Synthesis session around AI Search, Advanced Project Search is letting us take even more advantage of the knowledge that we have access to in the platform to answer questions about, you know, how many projects have we built in North Carolina in the last 10 years, or what are projects on where we've partnered with this structural engineer or this mechanical engineer, or what projects have Susan and Sarah worked on together? Like all these range of questions that come up all the time. That's what Advanced Project Search is going to help you with. So that'll happen between now and 2026. Between now and 2026, we'll also integrate AI search and the LMS. So you'll be able to search all that LMS content from courses and lessons and guides. Sorry, learning paths directly in AI Search. So that'll all happen by the end of this year. Which brings me to knowledge agents. So this is 2026 territory. We're in design now. So I want to share some of the vision of what we're thinking about with knowledge agents in the future of the synthesis platform. So how many of you heard the term agent kind of flying around, Just show of hands. Okay, that's a lot. Okay, keep your hands up. Keep your hands up. So how many of you, if I called you right now, could define what AI agents are for me? More than I thought. That's pretty good. Okay, there is no universe. You can put them down now, Todd. There is no universal definition. In fact, in the nerdy circles that I oftentimes spend my time in, there is a lot of religious war about, what's an agent, what's not an agent, what's agentic? How is agentic different than agent? It gets. You can imagine. So for me, it's pretty simple. An agent can take knowledge, reasoning, and tools to help you accomplish a task, answer a question, and really find what you're looking for. And so in the future, I think we'll be having a hybrid workforce of human employees and agentic employees. And they'll work together and they'll collaborate. I agree with the broad sentiment in this conference that we're not looking for agents to replace people. We're looking for us to really kind of upskill us, right? To take us to the next level, let us do more than we would do otherwise. So imagine kind of in your mind, just imagine an employee directory, but imagine it's a directory for full of these agents. They have different skills. Maybe they've got little avatars or thumbnails. They've got job descriptions, like they are meant to do certain things, like to really be part of your overall kind of company. So I want to share some kind of agent stories with you on what I've seen over the last year that really inspired us to build this into synthesis. One is a very personal example. As I mentioned, I've been doing more LinkedIn posting, and one part of the LinkedIn posting I've been doing, or one type of post I've been writing is sharing clips from podcasts. So I've gone on a lot of podcasts. This year we've been doing the podcast series with Evan. And so once or twice a week, I'll share a three to five minute clip from a podcast. And I've been using AI ChatGPT to help me write those little posts that go alongside the captions. I'm sorry, the clips. So when I started doing it, I love voice mode. Todd and I share this. I use voice mode like crazy. And so I start early in the morning and I'll be like, all right, I'll upload the transcript and I'll be like, I want you to write a LinkedIn post. I want it to have a really good hook, probably a question. I want you to stick to my language. I want it to be in paragraphs, like essay form. Don't do that little quippy marketing thing that you like to write. Stick to the structure and the chronology of my argument. Close it with a call to action that points people back to the full podcast, right? And so often every week I'm like dumping this context in so that it could do its job correctly. And then I have to go back and post it. So I felt good about that, you know, like that's it's got my good thinking already in the clip. I just want its help in writing this thing so I could offload some of that cognitive work to it. And it was working pretty well. But I then kind of like came across this process, like, well, I can build a custom GPT for that, right? And if you haven't used that, I'm using this as a very simple version of an agent. And Todd showed some examples of these as well. So I can just write that description once, that thing I keep telling it over and over to do, right? Those instructions that I'm giving it and I can build it into an agent and I can say all those things I just said. And as I learn new things about my process, I just update that job description. So instead of trying to recall it or kind of paste in this prompt, I just create this little agent, I click a button, I upload the transcript, and it writes a first draft for me. And it's somewhere between 60 to 90% good, right? And then I can just go back and forth and clean it up and make it better. So. So that's like a super simple, simple, simple version. Susan and I were on the road in April. We were in Austin, Texas. We went and visited a prospect and he had built a little thing he called RFP Buddy. So RFP Buddy chained together three different pieces of technology. One of the pieces of technology stored the past proposals and the boilerplate language and project sheets and a bunch of stuff. There was a middle layer that was kind of the agent layer that took the knowledge and connected it to the front end. And then there was a front end where you actually asked the question. So it was like super complicated, but you could see the value proposition. And so they go to the, they go to RFP Buddy and they ask questions. And what was good about RFP Buddy is it's already been just constrained just to that subset of knowledge, right? It was told like, how we want you to answer questions. Here's our brand voice. We want it to be whatever it is, paragraphs, bullets, there's like different formats that they kind of set up. And so it was kind of like a pre baked agent to help them answer RFP questions. So that's Cool. I'm looking at this, I'm like so you can kind of do that today with synthesis AI search, you can drop a kind of a detailed prompt in, you can apply some filters after the fact to narrow it down. But it's a little bit cumbersome and it's more of a power user thing. And so the idea of a knowledge agent would be bake that process into a really simple like one click operation for average human beings. So I think you can start to see these use cases explode. Like we've seen people trying to create these really. And this is mostly us seeing all of the search ratings and the feedback that's come to us. And I see some like amazing stuff coming out of our community. Like these super long queries like they're not searches, they are using AI search but they're trying to prompt it, you know, like it's an agent already. These super long queries around like right me a blog post, I'm not going to even do the whole thing but like write me a blog post about our community engagement work over the last two years and it will just use the content on the Internet, whether that's video, it's a post, whatever's happened. Or write me a blog post about our approach to mass timber. And then there's some filtering and they're kind of hacking the tools that they have to kind of create this experience where it can help you write. In our IT and DT folks, we've seen a lot of people once they kind of understood what AI search could do, they they're putting more and more of their internal documentation for their teams into synthesis, but then also their user facing documentation into synthesis so that people can self serve. But I've started to hear questions coming from the community. It's like what I really, really wanna do is I wanna get a help desk ticket, I wanna paste that in and I wanna have it draft me an answer based on the knowledge that we have. Right. And our style and the way that we wanna write answers. Awesome. Use case for an agent. Jess showed Purcell showed some things that she's been doing at Shepley, but others have too around kind of a video operations type agent. So the current workflow is you copy the paste the video name right and then you kind of copy and paste a prompt and say break this thing into chapters or write me a description or write me, help me break it into segments that I can make into lessons or write me a quiz. Right. That obviously seems like a knowledge agent that we could give a job Description. Say you are the quiz writing agent, or you are the chapter writing agent, or whatever it is. So again, we have a great foundation with AI Search to be able to package it up and deliver this experience to other people. The last one that I've heard come up from a couple firms is around hr. And it's this kind of idea of a seasonal agent around open enrollment. So imagine it's open enrollment for 2025. It's the end of the year. You've got whatever your source of truth is. Maybe it's a PDF, maybe it's a guide you've built. So you build a knowledge agent. You point it just at that guide. You tell it to be literal, like, do not be creative, like verbatim, like repeat content directly from that source that you give it. You put it on the homepage of your intranet, your knowledge base. Right. And then you point people to that during open enrollment as being the first line of Support. It's open 24 7. It can take a lot of the easy questions that HR gets bombarded with during open enrollment, whatever that two to four week period is. And then they can go to HR for the more advanced, nuanced questions or something that's not kind of covered in the default. And then in 2020. You like this, Angela? Okay. Yeah. Yes. All right. Good, good. I thought this would land. I don't know. I was just taking a risk. And then you shut it down, right? It's a seasonal agent. So at the end of open enrollment, delete the agent. And then in 2026, you stand it up again. So these are the kind. There are many more ideas. I'm sure some of you have written some down already. But this is the kind of, like, thing that we think that we're going to be able to do with knowledge agents. So a couple things that I think make these really important. One, I wrote a post a few weeks ago about the kind of AI literacy and what are our expectations and the kind of. The theme I tried to explore. Explore was, is it true that we think that we need to level up all of your employees to be AI All Stars? Right. In the same way. Right. And so, like, is that true? Right. And the things I started pulling apart were, well, let's look at BIM and the transition from CAD to bim. Is it true that everyone in your firm is a BIM All Star? They are not. We know that it's spiky, right? There's different levels of competence within bim, within your firm. In fact, it's not monolithic. Right. The things that project managers need to do is different than what a job captain needs to know. It's different than what an interior designer needs to know. Know. And I believe the same thing is going to be true with AI. AI is also not monolithic. We've got text tools, we've got image generating tools, there's workflow tools. So people are going to develop different competencies in AI for different reasons. But what I think this can unlock is even deeper than that. So if you're in a marketing team, for example, and you've got somebody on your team who just goes nuts with AI, right. They're in, they're early adopters of the technology. They, they're really good at writing complex prompts. They pick up our knowledge agent, they understand how to develop it. What that's gonna unlock is for the ability for the rest of the marketing team to take advantage of that knowledge and be able to work at that level or close to that level. Right. It will dramatically kind of democratize advanced AI capabilities by packaging it up in an agent. So you can just point that person at RFP Buddy or LinkedIn Writing Tool or whatever it is, and they can work same thing within it. Right. You can build an agent for doing help desk ticketing. Right. Same thing for video operations. Somebody doesn't have to understand advanced prompting. They can just use the agent and take advantage of some of that baked in technology. So I think, to me, we've heard since the beginning of really AI coming into the industry that we're like, we need to upskill everybody to really understand advanced prompt writing. And my kind of perspective as a product person has been I feel like we can design products that really kind of eliminate that from having to happen. I think we can meet people where they are. This is why when we built AI search, having a really good keyword search like we didn't, we kind of optimized for that first. It's like that they type in one word or two words or three words. We still want them to have a great experience. We don't want them to have to learn how to prompt in order to find what they're looking for. Now, it is also true that if you provide more context, you're going to get a better answer. But we still wanted to help people find their way into the tools. And I believe agents is another way to go even further with that. First to really kind of like say, I have a task, I want to accomplish it. Let me find this agent and let me do what I'M trying to do. So I think that's great. I was also in a conversation at lunch today actually and we were talking about. We didn't actually talk about agents. I was just listening to this company who. We were talking about their accounting team and how they have turnover recently and a lot of the turnover in their accounting team. And what's going on in my mind is, wow, this seems like a really good kind of risk management for whatever it is. It could be accounting, it could be marketing, it could be it. If you can transfer knowledge and make it available to an agent, that's a really good line of defense. When somebody leaves the company or somebody's on vacation or somebody's on family leave or in Gillian's case. Right. It's a global firm, so maybe the expert's asleep. Can we build an agent that could answer at least basic questions and get them the knowledge they need? You know, while the. While the technical person gets to sleep and doesn't get woken up in the middle of the night. Hopefully this making sense. So at its fundamentals, I mean, we're just getting into design. I don't have any slides for you, but the key kind of idea is you'll create an agent, you'll give it a name. We don't have figured out the thumbnail. We haven't figured out how far to personify this thing. So it might be called like an avatar. I don't middle ground like thumbnail feels a little cold. Profile picture feels a little too humanish. So maybe there's something in between. So you'll give it some kind of image and give it a job description and that's the language. So you are an agent that does xyz. When you get these kind of questions, here's how we want you to write. You'll connect it to a certain specific set of knowledge. So it could be a guide, it could be a library, it could be an LMS course or a learning or whatever it is, you can kind of constrain its knowledge and you're basically telling it, here's what you do. We believe these knowledge agents need to be available to individuals to create their own. So their own personal knowledge agents for teams. So I can create a knowledge agent and shirt within the marketing team or for all employees. So maybe in the self service example during open enrollment. That's a good example. Another kind of key design principle for us in version 1 of Knowledge Agents is human in the loop. So if you kind of tap into the broader discourse around agents, a lot of the Focus is on these totally autonomous, asynchronous agents that are taking action on your behalf, or maybe not even on your behalf. They're just like fully running, you know, kind of parts of your organization. That's not what we're trying to do in version one. We're gonna build them a Human in the Loop. And what that means is anything that an agent does will be initiated by you. Right. So you'll go to it and say, like, help me with this proposal. Help. Help me write this help desk ticket. Help me with this video. And then you will then also take the action. So you'll write the blog post, you'll send the email. It's not gonna take action on your behalf, and it's not gonna run autonomously or all the time. That's not to say we won't ever go in that direction. But we have to get through Human in the Loop. That's a great place for us to start, right, to build these ones. We're Human in the Loop. It's a way less risky way to get into this technology for you and for us. Right. And as we go through Human in the Loop, use cases where you have to interact with the agent. Then we'll explore. Maybe some really obvious opportunities will come for actions we can take. Right. Tasks that can be completed, emails that can be sent, whatever the thing is. And also use cases might emerge for them to operate independently. But that is kind of more science fiction. That is not a 2026 thing. And so if we're back to, like, I'm talking about what happens between. By the way, we're still on what happens between now and 2026. I'm just being cake in at 2026. This is what we're looking at. So that's kind of the kind of the big picture. You know what we're thinking about with knowledge agents. Is this making sense? All right, cool. Do we like this idea? Okay, so I told you, we're going to do three things. We're going to talk about LMS and learning and development. We're going to talk about AI and then we're going to talk about it coming together to do that. I want to kind of go to Hollywood film history, and I want to ask the audience who the most famous agent in film is. Ok. Sundance Kid, let's say sports agent Jerry Maguire. Thank you. All right, and so for Jerry Maguire, what was the famous advice or kind of plea that that he gave to his client? There's another one. Help me help you. That's it. So Jerry Maguire, the sports agent, says to his client, help me help you, and knowledge agents say to you, help me help you. Right. If you want a knowledge agent to be very helpful, you're gonna have to help it become effective. Right. And this is why we're gonna need to be doing knowledge management and learning and development for agents as well as humans. Right. In order for an RFP buddy to help you, it has to have the knowledge it needs to get its job, and it has to have a clear job description. In order for a video agent to help you, it has to have access to good quality knowledge. In order for an IT or a DT support assistant to help, what does it need? Access to? Good knowledge. Right. So we need to be thinking about learning and development for these agents as well as our humans. And the good news is, most of the time, it's the same thing. Right? Right. It's the same knowledge. The same knowledge that's gonna help your agents do its work, will help your employees do their work. And Ellen and her team talked in their breakout around kind of writing at the level, kind of an intern or emerging professional. Right. Clear, specific, define your jargon and your acronyms, provide context, including the why. That's really good writing to do for employees, but that's also really good writing to do for agents. Right. So help me help you. I hope that sticks for you because that's gonna be a thing we're gonna be talking about a lot. So let's go forward. That's probably good enough for K Connect 2026, some LMS stuff, some AI search stuff, and some knowledge agents. So let's go forward to K Connect 2030. I think that more is gonna happen, more change is gonna happen in AEC knowledge and learning management in the next five years than has happened in the last 25. That's my view. Between now and K Connect 2030, more we didn't have K Connect 2000, but play along with me. If we had had it in between now, I really think the next five years are gonna be transformative, and I think that's generally a good thing. I think we're gonna create learning experiences for your employees. I think we're gonna get them the knowledge that they need in the time that they need it when much better than we ever had. I think this is a dream era for knowledge management people to be working in. I think it was Earl that said earlier, like, you're lucky that you're coming on board now with Synthesis six. And I take it a step higher, and I'd say you're lucky you're coming on board to AEC Knowledge Management, learning management in 2025, because I think it's going to be a heck of a ride. Some of us who have been doing it for longer, you know, we'll probably be crusty in the back being like, oh my God, do you remember what this was like in 2005? Like it was so damn hard. Or do you remember what this was like last fall when we had keyword search? You know, like it's really, it's happened quickly, I'm going to tell you. So I think more change is going to happen in the next five than has happened in the last 25. And the reason I'm really confident that we're going to do it well together as a community is because we've already done it together. And by that I mean I've been saying a lot lately. I feel like it's 2009 all over again. When we first got started as a company, when we first got started, we were trying to bring what I call now, KM 2.0 into the market, right? So KM 2.0 was about bringing the social capabilities of tools like Facebook, LinkedIn and Twitter inside the company for knowledge sharing. And it opened up all kinds of possibilities, right? All of a sudden I could ask a question and I'd get an answer from across my company, from people I didn't know, cutting across silos, offices, departments, whatever it is, I could communicate and share what I was learning, lessons learned, New technology. It was just for those of you that weren't around then, this might sound really quaint, but it was a big deal because before KM 2.0 everything was pretty much top down. Everything was approved, it was kind of pushed out and bringing that new technology into intranets. We even called it a social intranet at the time. It was really a big thing. It was not universally loved. This idea, for those of you that remember this era, there was a lot of fear like what happens if people post the wrong thing? What happens if they say the wrong thing? What if they give someone a bad example? Are we going to be able to be sued because of some comment on some post? Those questions went on for multiple years and I've said multiple times over this conference, other people have said it too. We talk about knowledge management being about people, process, technology and culture. The people, process and culture piece, that's what we've been doing at K Connect for the last 15 years, is really figuring out all that other stuff you have to do besides just adopting new technology. So in that example, km 2.0 social was the transformative catalyst. And now that we're in km 3.0, the AI is the transformative catalyst. And we know as a community, we can figure out all that people, process and culture stuff together, including the transition to being more modern learning organizations. Because we've done it before, right? When we first got started in 2009, we didn't have essential content reviews, we didn't have essential content checklists, we didn't have information architecture, we didn't have a design inspiration gallery, we didn't have all the intranet tours, we didn't have a K Connect Talk archive. We didn't have any of that stuff. And we built it together as a community. KA was here to help orchestrate it, but you guys are the people that filled it in. We designed the stadium, but you're the athletes. We designed the music hall, but you're the musicians, right? That's what's made this stuff all worked. And none of us had all the answers, but we worked together to build it. So now that we know what that model looks like and what it means to build those best practices, what it means to build community, we can go a lot faster in this km 3.0 transition. So as we get into adopting learning and development and trying to build more modern learning organizations, we know that we'll do that as a group and we can make it through. So I think that it's a really exciting time. I am so grateful that we have this community together. I think I've said the phrase getting smarter together a bunch during this event, and that's how I feel. Right? That's how I feel. As we go forward, we're excited to be able to build technology, but know that that technology is going to do amazing things in people's hands. To illustrate that, I just came from the Future of Synthesis LMS session, and I watched Andrea and I watched Amy and I watched Marco talk about what they were doing with a tool that just came out five weeks ago. Seriously, five weeks ago. And their level of rigor and thinking around philosophy and change management and strategy and aligning priorities like it was. Like it was lit. I know the kids don't like when we say lit these days, but I still like it, so I'm gonna do it. And so that was after five weeks of seven firms having this tool in their hands. I was looking at this, I'm like, this is some really advanced thinking in what we're doing. So imagine what it's going to look like when we're back here next year or we're here in 20, 30. So I want to thank all of you for being here. I want to thank you for being such a great community. And let's go get it.