KM 3.0: A Golden Era for AEC Knowledge Management

Welcome to the Troxell Podcast. I'm Evan Troxell and in this episode I welcome Christopher Parsons. Chris is the founder and CEO of Knowledge Architecture, a software company that's been working at the intersection of knowledge management, community and technology in the AEC industry for over 15 years. He's also the creator of Kaconnect, a long running knowledge management conference and community for AEC professionals. In this conversation we demystify what knowledge management KM for short, actually means in our industry. Chris walks us through the evolution of KM, through the early days of knowledge capture, which is KM 1.0, to knowledge sharing KM 2.0 and into today's emerging era of knowledge in the flow of work, Otherwise known as KM 3.0. We talk about why this shift is happening now, largely thanks to the rise of AI and and its ability to deliver relevant information when and where it's needed and why tools alone aren't enough. Cultural process and operational transformation are the real unlocks here. As Christopher says, knowledge is infrastructure and firms that want to evolve must treat it that way. As a bonus, you'll hear real examples of how firms are already creating value by operationalizing knowledge. And just as importantly, we dive into the human side of all this how language, metaphors and cross disciplinary communication play an essential role in making knowledge work for everyone, not just a select few. A key theme from this conversation, which connects to many other episodes here on the podcast, is that operationalizing knowledge is a team sport. It's no longer enough to expect a single person or department to steward institutional memory. In the KM 3.0 era, everyone in the firm, designers, technologists, operations leaders and executives have a role to play in capturing, structuring and sharing knowledge. And when it works, it unlocks not just better decisions, but better culture, better retention, and better outcomes for everyone. As always, there's an extensive amount of additional information in the show notes, so be sure to check that out and you can find that directly in your podcast app. If you're a paid supporter of Troxel and and if you're a free member, you can find them at the website, which is again trxl. Co. Lastly, you can really help the podcast by sharing these episodes with your colleagues and by commenting and sharing on my LinkedIn posts. You can also leave a comment over on YouTube to engage with me and other listeners. And now, without further ado, I bring you my conversation with Christopher Parsons. So Chris Parsons, welcome to the podcast. Great to have you on this one for the first time. We've we've talked many times over the years. You've been on other podcasts and episodes, and you've also suggested guests for this podcast. So that's been fantastic. So welcome. Officially.

We made it.

We made it. You've finally gotten there, Chris. I've just been waiting for this one box to get checked. And you finally did it. I can't tell you what it is.

I guess you made it. Now that's what we should be saying. There we go.

Well, I appreciate you coming on and talking about knowledge management today because this is a subject that's near and dear to my heart. Maybe before we jump into what that even means, tell us, like, what's your story with within aec?

Yeah, I was a history major in college and graduated in the middle of the dot com boom, which for I don't know if all your viewers know what the dot com boom was, but our listeners, it was just explosion when the Internet took off and everyone was digitizing their businesses and building dot coms. And if you were. I mean, I graduated the history major, not a CS major, but if you had a heartbeat and could do anything with technology, you had a job in, in tech back then. And so I was a technology consultant, went around the country working on these projects where we were laying, you know, networks and infrastructure and like accounting systems in for the first time. Like, it was kind of a wild thing to imagine, like putting those first systems in. But that's what my job was. And moved to San Francisco in 2001. And then later that year, or sorry, 2000 doc, the dot com crash happened and 911 happened and I was eventually, after four rounds, you know, laid off and started looking for work in the city. And my kind of constraint that I said is I'm like, I want to. I moved to San Francisco. So I sold my car and I wanted to be able to walk to work. And I'm like, okay, so my constraint is I'm. I'm going to. I'm going to find a job within the city of San Francisco that I can walk to. And in that time, tech was like dead. And you could only get a job in tech if you had like 10 or 12 years of experience, took a massive pay cut. Like, it was terrible. So like for someone two years out of college, like, that wasn't happening. So I found a job in this architecture firm, smwm, to ride out the recession and then go right back into tech. That was my plan. I didn't know that architecture firms existed in the form of more than five people. You know, I didn't know that there were like fifty hundred thousand person architecture firms. Like that had just never been part of my life. And I just really love this company. Like I love the intersection of art and architecture and science. And we did a lot of public work. So there was like politics and kind of public policy involved and the engineers, the collaboration that would happen. I just, I'd go to all of our crits, I go to all of our kind of, we call them forums but lunch and learns, soaking it up. Taught myself a lot. I went to a lot of classes because I wanted to Revit came out during that run. And so I like, I wanted to help the firm with the transition to bim. Actually we tried archicad first, which is a whole nother story.

I taught archicad before I taught Revit too. So I'm with you.

You've been there. Yeah, it was fun actually. So I did that for five years. So my kind of plan was do this for a minute and then get out. But like I actually loved it. And then went to another architecture firm, a larger one, Steinberg Architects became their chief information officer. And for about eight years I worked those two jobs. And in each of those companies I, number one, built intranets and it was a lot harder than it should have been. And number two worked on knowledge management. And so I kind of discovered that in my first firm I'd kind of built the kind of core IT stuff, servers and printers and email, like the stuff you could do in any company. We had design technology going. We didn't call it that at the time but like it was becoming digital design. Like we didn't know what the terms were going to be.

Yeah.

But then this knowledge management thing. So intranet, but a CRM and image management and like all this kind of information systemy stuff, but as well as learning and development, like that's where my heart was. Like I thought that was the most interesting part of my job. And so after having done that at two companies, I started looking. Well, I can either go to a large company like the Genslers and Soms and Hoks of the world and like run knowledge management there and just only do km or I can start a business and work with kind of more small and midsize firms that I'd grown up in and help them with knowledge management. And so did that in 2009 and I didn't know honestly what that was going to mean. You know, like, sure, it's, it's so clear now. Like that synthesis, our Internet platform was going to be the thing. But I, you know, there was, that was some other integration software. I thought I could maybe be a virtual knowledge officer, like one day a week for these firms and get four clients and take Fridays off and send bills, whatever it is. But it became click within the first six to 10 months. It was like, okay, we're going to do intranets. Like that's what people really want. Some people outside of our company had seen what we built, what I built before and said like if you could build something like that as a product, like we would love, like we've tried to make SharePoint work and it was hard and we'd love to hire you to come in and help us do intranet. So yeah, that was 2009 and now we work with over 150 firms across the country. We're on version six of the platform. But I think more critically, we kind of realized early on that knowledge management was more than software. Like it's people, it's process, it's culture. And so we believe that in order to have the kind of impact on the industry we wanted to, in terms of bringing knowledge management to AEC and having people see success, we had to focus on that other stuff, people, process and culture. And so started our conference K Connect in 2010 to bring together, you know, we had 30 something people speak at the first one, but they were short talks because that's kind of like when there's a new thing happening and changing, everybody's got like a third little tiny piece of it figured out. And so the best thing we know how to do is create community to help bring those people together and share. And so that community and the sharing ethos of community. KA has been there since day one because we really wanted to do more than just be a software company. We really wanted to help lead this movement in the industry. So that's the kind of high level from 99 to 2025 that I, that I can kind of tell you about.

So many things went through my mind. Like when you're, when you're talking about spinning up networks and I'm thinking Novell Networks back in the day. And I'm, there's just so many things.

That we've been through and then, you.

Know, maybe you can define knowledge management. So, so a couple things, knowledge management, but then you're also saying ka, which is not, I just want to put it out there, Knowledge Architecture, right, That's the name of your company, you call it KA for short. And then the product that you're talking about, the Thing that you do is you make an Internet called synthesis. And I'm just thinking of kind of how this has evolved because firms. I just had a guest on the show recently talk about how like everything for a very long time has just been locked into fold. Not locked, but in folders somewhere on the P drive of the network. Right. It's just like, hey, where's that thing? And it's like, oh, it's on the p drive in 386244. And it's like. It's like really?

And.

And so this is. We. We've also seen social media come along where there has become even before that, right. We've. We've had LiveJournal and MySpace and Facebook and like this growth of what a social media platform is and that, like this idea of an online community which is kind of a mirror or. I don't know how you might describe it better than I can, but like this. It's this other version of your institution, your organization. Right. A lot of times and sometimes there's similarities, but then oftentimes it's like where people go to look for stuff. And that, like I said, stuff used to be in folders and that was really opaque. Right. It's hard to find stuff. And we're talking about. When you talk about knowledge management, I want to hear your definition. But these are all the things that I'm kind of thinking of when it comes to, like there's stuff locked in people's head, there's stuff locked in email, there's stuff locked in folders. And again, it's not locked. Like it's not behind that door that you can't get into, but you have to go looking for it and it's difficult. And so give us a definition of what knowledge management means. Maybe. And maybe it's evolved too. Maybe you can talk about different versions of what knowledge management has been. Yeah.

I'll start with a short definition and then maybe I'll kind of walk back and talk about the evolution, I think, about knowledge management, about connecting people to the right knowledge and the right people in the flow of work. And so when we first. So knowledge management really became like a thing, like a term that got. You'd find in Harvard Business Review or mentioned at conferences kind of in the 90s. But I mean, obviously we've been doing. We've been managing knowledge as a species for a lot longer than that. So, you know, going back to cave paintings and oral traditions and books and libraries and universities, like we've been trying to Figure out how to write down what we know and capture it and share it and get it to the right people. That's a human, very human thing to do and to build technology and build tools that help us do that faster. I mean, tool building, we share, architects and software people share that. Right. We like to build things. And so. But in the 90s, there, there was something that happened, was the beginning of that, of the Internet. And you saw a lot of that knowledge that had previously been very analog. Like, I don't know if you were in practice when there were still binders of like, oh yeah, you know, yeah. So the HR handbook, everything was filed in a binder. Everything was filed in a binder and file cabinets and it was all paper and hard to retrieve. And if you're working in multi office. Oh my gosh. So what you saw with the Internet is you saw this digitization of all of that stuff. And so you saw the rise of intranets, file servers, databases, all that kind of technology that basically took that knowledge that was captured and made it more available digitally, easier to, you know, kind of search easier to access. But it was very focused on knowledge capture. And it was very kind of top down in a way most of the time, right. It's like somebody at the senior level of the organization saying, this is how we do things here. And then the digitization became like an efficient way to like push that down. In 2007, this is when, you know, I kind of marked social media taking off. So Twitter, Facebook, LinkedIn, the like. This is when I was at that second firm and I started looking at that being like, you know, we could use that technology inside the company, right, for knowledge sharing, people asking questions across offices, across disciplines, sharing new things that they've learned or lessons learned or new technology, whatever it is, like we have a lot to share and putting it in a place beyond email that could be searched later or if you join the company after those discussions that happened, it's not like they were invisible to you, like you could capture them, but allow them to be more social in nature and richer with media and that kind of thing. So we started doing that when I was at the second firm and then that became the synthesis when we launched. It was a social intranet. So that was like, that was a very big idea at the time because intranets had always been very static and very. Yeah. And so we were having dynamic intranets and not only was it dynamic, but that it was democratic. And this was. Every time these technology shifts happen, there's a culture Shift that has to come along with it, right? And so I remember there was so much resistance to synthesis in some firms in the beginning because they're like, well, what do you mean? Anybody in the company can share, right? Like, this is terrifying, right?

Permissions, permissions, permissions, permissions.

But then I want to approve every comment before it goes up. And like, what if they say something wrong? What if they say something inappropriate? And what if they tell somebody about this product that we don't want them to use? Like, it was a lot of fear. And like, that seems kind of funny in 2025. Like, we figured it all out, but there was a lot of concern about that. And so I think that we rode that wave into business. And like, and this is right when K Connect started. So a lot of what we were talking about at our conference was about how leadership is modeling good sharing behavior or how to get people answers. Or, like, trust being so central to knowledge sharing. There was this. I don't hear this as much as I used to when we started, but people talked about knowledge is power and people hoarding knowledge to retain their power within the organization. Like, I don't know, I hadn't really thought about it until I just brought it up that, like, I don't hear that like I used to when we first got started. I think companies have maybe become more trusting in general, or people have understood that sharing and building others is a. Is a way to advance. I don't know. I. Yeah, I'm pausing on that one for a second. But that's a good progress. I mean, I think I'm happy about that outcome. And so, like, here we are now in 2025 with AI arriving and that's the next big disruptor. And every time there's been a tech disruptor, it has changed the way we manage knowledge. And so, you know, there's a lot of different ways that people are looking and we can dive into this if we like. But looking to use AI around knowledge management. But the big way that knowledge architecture has been thinking about it has been from AI search and retrieval. So we've captured all this good knowledge. People have been sharing knowledge, but now I want to find the thing. I've got a core question about a pipe, or I've got a question about a grab bar or a, A standard, or who's the person I go to being able to search across all that knowledge and either find the answer or find the person who can help me, mentor me. Like, whatever it is, worked on a certain project, like, that is now what we think is really, really possible, we're seeing that in our, in our technology. And it's really exciting because it's always been the hard part is like the cat, you know, to get at the answer. So that's kind of how we're thinking about knowledge management now. I named the company Knowledge Architecture because the firm I worked at, the first one, both firms, but really the first one did a lot of urban planning and master plan and I always liked their presentations and I loved the kind of long view that they took around. You know, we have to think about the whole campus strategy and then it's going to be this building, then we're going to renovate that thing and then this group can move over to that building. Like the logistics of it were super interesting.

Yeah.

And I saw in a massive parallel to how I thought about knowledge management, you know, like you can't, there's so much knowledge in your organization, you can't manage it all at once. And so you have to think and see, you have to prioritize and think in terms of like moves that you need to make. And that could be in terms of systems too. So like when we, you know, talk to a prospective client, like a lot of times we find ourselves saying we might be the right fit for you in three years because it sounds like you need to do X, Y and Z to get things in place before it makes sense to bring us in. Yeah. So that, that longer term knowledge architecture, like thinking about literally the architecture of systems and how knowledge flows within your organization like it's three dimensional chess, like I think in a lot of ways. So that, that's kind of how we got the name. But that's, and that's how we think about it too. It's not like there's one monolithic system that's going to have everything. It's really a dance between systems.

One of the things that I keep thinking about, having gone through an intranet implementation previously too is like, I keep thinking of the word participation. Right. And because you talk about kind of the difficulty that it's to get to this point where now it's, it's what's incredible with, with tools like AI is like practically everything is searchable at this point, which has been difficult. Right. To say the least. Previously when, when you've got a, a giant BIM model, when you've got a long video, when you've got a long audio recording, you've got threads upon, threads upon, threads of email tossed in there along with all the other Spam and all the other stuff. It's like, how do you find the needle in the haystack, right? Because to find the right thing in the flow of work, like you said, it's like, okay, well, with the least pain possible, with the least friction is kind of this magical thing. It's like that to me is what is so interesting. Like, like just the parallel of, oh, if I go to Google and I search for something and I see a ton of ads or a ton of placed links, right? I have, now I have to like, Google used to be so dang good at this kind of thing and now it's subpar, right? It's, it's not a good experience, right? And, and now I go to chat GPT and I type in a thing and it just tells me like, it doesn't try to sell me something along the way. It's not putting, you know, something adjacent to a product in the way of me finding the information I need. Maybe that'll change in the future, I don't know. But, but just as a, as a analog to what you're talking about, it's like, well, the least friction possible is often what I need to stay in the flow of work. Because the detours cost a lot, right? This is one of the things where it's. When we used to talk about it when we were putting in the Internet, it's like, well, do you want to go to the one place the single source of truth? Or you know, because when, when we go out looking for something on the Internet, it's like all of a sudden I'm buying shoes that I didn't know I needed, right? Because I got distracted along the way. And that, that's, that's a real thing, right? But I do come back to this idea of participation and a behavioral shift to get people to. Because you're talking about like, what do I need to do to find the thing about the pipe or whatever. And it's like, well, somebody needs to put that information in there. And now they don't have to explicitly put that information in there, right? I think one of the cool shifts in the latest version that what you're talking about of knowledge management is like being able to find the needle in the haystack. That could be a two hour long interview with a principal about everything that they know about some ANSI thing or some code requirement. And to me that's super amazing that that can happen now. And the friction to participate is lower because we don't have to whittle these pieces out manually and file them in the right place and do all those kinds of things. I mean, what do you think just generally about, like, the way things have kind of evolved in the. How we store information, but also how we can access that information to make it less friction to find the things that people need when they need it.

Yeah, yeah, there's a lot I like in there. I mean, there's so many things that we've been doing in knowledge management. I don't just mean at KA and even within aec. I just mean knowledge management in general. You know, like some of the greatest hits, Lessons learned over the years. You know, there's this notion of critical knowledge transfer which you were hinting at, or expert interviews that were really, really good ideas. You know, they recognized, you know, let's take lessons learned. They recognized that, like, it'd be good if our firm didn't have to keep learning the same lesson over and over again. And the next time we, you know, encounter this condition or this challenge on a project, we can learn from our past projects that we don't have to just learn the same thing over and over again. The problem was operationalizing those insights. So somebody writes it down and what puts it in a lessons learned database. And now what's the next person supposed to do? Like, think of all the potential things that could go wrong. Search this lessons learned database on, you know, like, it's just, it's just kind of wild. Like, so what you had to do back then is like, the only way to get lessons learned useful is to operationalize them and like change policies, change standards, change training, you know, whatever it is, you had to really put them. And that's still a good idea, don't get me wrong. But I mean, even the lessons learned database now has gotten some new life breathed into it because now somebody doesn't have to think. Like, the way that we're approaching it at Knowledge Architecture is like, I don't have to think whether it's in a lessons learned database or if it was in a video or maybe it's in the standard section of our intranet. I just ask my question and, you know, AI will be smart enough to retrieve, you know, the right answer whether it comes from lessons learned or not. And so, like, I think it's allowed, like, to your point, like, it's removed a lot of friction from the content creation process. Same thing with interviewing folks to transfer knowledge. Like, it's really hard for experts to write down what they know for two reasons. One, they're super busy and they're expensive and billable, and finding time is hard. Writing is hard for them. But also, most experts don't know what they know anyway. They don't know what it is that makes them special. Like, it's what we call tacit knowledge in knowledge management. It's silent knowledge. Right? So the two types of knowledge are explicit knowledge and tacit knowledge. Generally, it's how people talk about it. Explicit is stuff that you can write down. It's standards, procedures. This is. These are our rules. Like, this is how we do things. But tacit knowledge is like, I don't know. You imagine if you ever worked with a CFO that can look at a gnarly spreadsheet and, like, point to the one cell. They're like, it's because that's off, you know? Or like, you imagine a pilot who can fly a 747. And if you've ever seen the cockpit of one of those things, it's like thousands of switches and they know exactly what to do. It's all like compressed intuition. Like, they don't. They don't think like, oh, I should go find this button and do the thing. It's like they just get a feeling that something's off. Or like, this is what needs to happen. So it's hard if you ask then an expert to write down what they know. It's very hard for them. So interviewing them is a really good technique. But then the problem was always on retrieval. So it's like, how do you then take those insights? So you imagine somebody searching those past videos using keyword searches or lessons learned using keyword searches. And they get seven videos, 10 videos that are an hour long each. What are they supposed to do? Watch all seven or ten of them. And it's not realistic. So it's like a break. It's like we kind of have stacks that capture the sharing. And now we're getting this flow of work thing that's the big breakthrough of the unlock. That's pretty exciting. I think I answered your question. Oh, I will say this, the participation part. So now here's what we're seeing is that experts kind of. They knew the trick. They kind of know that it was hard for people to retrieve. So they're kind of a little bit dragged their feet to participate in that kind of stuff. They're like, I know it's good for the company. It's definitely good for the person interviewing me. They're going to learn something. But, like, how does this actually, like, three years from now, like, is this getting used, I don't know. Now they know it can be used. And so now they're like, okay, I see this as leverage. So if I'm a subject matter expert and I want to make, I want to, I don't want to be the subject matter the only one. I don't want to be the bottleneck. I don't want to have to answer the same question over and over and over again. I want to empower people. If I'm at a job site, if I'm with a client, if I'm on vacation, if I'm at a conference, like, I want people to be able to work and tap into the things I've learned and I want to like, almost like, I want to like free up that space in my brain. Yeah. That's holding that knowledge to make room for new things so I can, you know, kind of continue to lead and evolve and grow and strategize. So I feel like for subject matter, for both people looking for information, like seekers and subject matter experts, this is a really interesting time because I think both of their quality of life, quality of lives, quality of lives are about to get better.

Yeah, that's a really good point. And, and, and so this brings up something I thought about a lot, which is people need to kind of see it to say, oh, okay, it's possible. Right. Like they actually need to see the case study to give themselves permission to do the thing. And I think a lot, I see that a lot in architecture. It's like, I mean, there's always talk about, oh, the hourly business models working for time is we can't, it's like, well, show some other successful examples and then people will give themselves permission to pursue it or incubate a startup, technology startup inside your firm. Well, show me another firm that's successfully done that. Right. And also talk about the failures along the way so that people know what they're getting into. I mean there's a lot of different types of examples, but specifically with knowledge capture, it's like, I do feel like a lot of people think, oh, what, you're just like, what is this? This document is going to get captured and then no one's ever going to use it. But that really is now very possible to show. Like, no, actually look like we can act. Let's just do a quick 10 minute one or let's just show you that one that we did with somebody else. And the tools are there, I mean, to actually accomplish what you're talking about and you've shown some Great examples recently that I got to see. So it's.

What's interesting about what you're saying is it goes a step further than that. Because if you now the kind of. The new rules of the game are if you capture really good insights, they're going to get used in ways you couldn't imagine when you captured them or when you shared them as an example. You know, a lot of people record their town halls or learning and development sessions or lunch and learns or whatever it is in our community, and they upload those into their intranet. Meaning. Meaning for them to be continuing education. But the marketing and communications teams will want to ask questions like what's our approach to mass timber? And they might be getting results from a lunch and learn a technical debris that wasn't there. They weren't the audience. Right. But they're trying to describe to the marketplace how the company thinks about a topic or an issue. And it's drawing on all that knowledge that was transferred more for professional development and continuing education. And it's helping them craft a narrative on how the company thinks about something. So this idea that it can then synthesize knowledge from like multiple different places to help you get kind of a unique answer to your question. Yeah, that's also brand new. Like, there's just. We're in the first inning of this thing, Evan. Like, we're just really trying to figure out, like, all the ways it's going to help people just build smarter firms.

Is that 3D chess, 4D chess, what.

It is, like, it's connecting things on.

Different dimensions and different vectors. That's really interesting and I love that point because I think about how AI is unlocking this and maybe even more specifically towards AEC firms in this regard. Because one of the things that you've talked about in the past is I don't know if you're the first person to bring this to my attention, but you've definitely made it more of a point that I need to pay attention to, which is this idea of rag retrieval, augmented generation. Is that the right.

You got it.

So good job. I get a gold star. But the thing is, can you talk about what that is and why it's important here, especially around, like, vocabulary and jargon within the industry, because I think a lot of times people are using AI tools and you might search for something and it finds something, but it's not the thing you were looking for because it's not specific to architecture. But I think we've also seen it with like transcription services. Right. Where it thinks you said something else than you actually said because it's not trained for your industry. Is that. Am I in the right ballpark?

Yeah, I think there's a few things that are interrelated there. So retrieval augmented generation. When, when we first decided to build an AI search product, we were looking at. So the way that OpenAI kind of pre trains their models, I think most people know this by now, so I won't go into depth, but basically it's a pre trained model on content that was that content at the snapshot of time in which that model was trained. So when the first, I mean you might remember when 3.5 or whatever came out like the data was 8 months old.

Yep.

So that was a problem.

Yeah.

If you were going to try and use that for the use case that we're talking about. In addition, if you were trying to train a model, there's all these permissions issues. So principals should be able to see some things that maybe interns shouldn't. So training a model wasn't good for that. And it was a hallucination issue like basically, you know too. So RAG came out as kind of an architecture which basically said, look inside the company. We're going to do this another way. We're going to do a search first. So that's the retrieval in retrieval augmented generation. And we're going to pull snippets of like the best content that matches the query intent behind that query. Whether it comes from a document, a page, a video, an employee profile, doesn't matter. And then we'll set. This is when we'll use AI is to kind of summarize. All right, what should. Now here's the query. Here was the intent. Here is the relevant content that we have in our company that could answer that AI. Please stitch this together into a summary and then include citations to let people have reference ability back to where that information came from. So that was like a really transformative, you know, for knowledge management. Like that was really kind of the only way to do this. Trying to train a model like for an internal use case just is problematic on 100 levels. Not to mention cost and effort and all those things. Right. So that's, that's RAG in a nutshell. And it's been really, really powerful. It turns out that the hard part about RAG is really not the AI part, it's the R. It's the retrieval. Like it's the retrieval, it's the retrieval, it's the retrieval. It's like it's so the challenge because it's also the data like it's, that's where the heart, you know, the hard, hard thing is. So we've, so when we've released 1.0 of AI search, it was meant to be just like a very general knowledge base search. So what's the best way to do this flashing thing? Or what's a revit container file and how do I use one or you know, all these kind of General. What's our 401k policy like? From benefits to design tech to all of it. What it wasn't good at was answering employee questions. So like how many licensed architects do we have in California? Who's the go to person on this subject? Because that kind of data was stored in a structured database. You know, those are like more structured database questions. And so we had to keep working with the retrieval part of RAG to understand the query intent and say like you know, if it's just questions about licenses that should be looking at a database. If it's a question more about like a more general knowledge based question, it should look to the general intranet or in some cases if it's an employee like skills, it might be both. Right. Some of that might be in a structured database, but it might be that Evan is listed as the contact on the revit container file page and maybe he's the go to person. So the hard part that we've been working on is not the summarization. That is a cool inherent property of these large language models. It's been getting the context and the content to that the right stuff, understanding the context of the question, understanding the resources that people have available within their kind of systems in order to find the right content to answer the question. So that's been the magical part. So now we're getting into agentic RAG where we're using, we're describing all the tools, right? All the content that it has. Here are some tools you have for retrieval. You AI make smart decisions based on this query to use the right tool to look in the right place, to bring the right answer back. So it's this kind of multi step looping process now versus a one shot. Here's, here's the, here's the content. Summarize it. So it's getting more sophisticated very quickly. Pretty interesting. I go too far. Is that, is that a good.

I think it's, it's not for this audience. It sounds super valuable to me too though that, that like you're connecting. Just call it all the departments in this way in a firm because firms have lots of, you know, I. Maybe silos, maybe that's not the right word, but. Right. There's these departments for sure. There's like, there's marketing, there's hr, there's Risk management, there's qa, qc, there's all these things. And rarely have they talked to each other unless you go directly to a person and then it's like the gatekeeper. Right. And then you schedule the meetings and you talk to the right person and, and that what you talked about a minute ago, kind of this unforeseen unlock of like tying these various things together with one query. And AI can go out into this lunch and learn. It can go over here into this town hall and it can go over into this somebody's State of the Union kind of a thing or whatever and pull something together out of all that. It's really cutting a cross section through all these different places. And I think what you're saying now is with rag, it's like it's actually determining the appropriate place to go look for that based on this kind of training data that it has. But again like super valuable to shortcut the line to get to where you're trying to go.

Right. And because we're domain specific with aec, we can give it a lot of shortcuts. Like if you had to try and get SharePoint to do what we're doing, it doesn't know the details about projects and roles. And I'll give you a really silly example out of the box. The AI couldn't figure out that registered architect and licensed architect are the same thing. Just like they.

Yeah.

And so people would search for license and they couldn't find it. And so like we're, you know, you build this little context in. We have to sometimes provide these contextual clues to help AI find its way too because like we're coming from a very industry specific problem. You got this interesting thing about silos because I think silos are an interesting thing because usually it's used pejoratively when people talk about silos. But they're also quite helpful in many ways. But where they become one place that they've always become really tricky and contentious is in information architecture. Meaning how do you organize the navigation of a website or an intranet? What are the broad buckets across the top? What goes at level one, level two, level three? Like how do you organize information to make it easy for somebody to navigate and find it? And you would get into political wars within companies. Again, website and Internet, it doesn't really Matter, like, it's just like, because you're basically asking very. Like, there's questions about power and hierarchy. Like, there's a lot of stuff baked into physical space.

Who gets the private office? Who gets the corner office?

It's the same. It's the same. And like, by putting me in this office space or by putting. At this level of our navigation, you are basically saying things about me.

It says something, right?

It says something. And what's interesting, like, I don't remember who our first client was. I think it was Hillary said, like, are we still going to need to care about information architecture in the AI age? Is this important? Because a lot of the distinctions that get made are artificial. Like, there are a lot of things even though, like, I'm trying to think, like, I mean, I don't know, like, proposals even. Like, that's kind of lumped in as a marketing thing. But is it like, principals are responsible for bringing work into the company proposals is one of the ways that they do it. You know, project. You know, like, project managers need to know about proposals. Like, it touches finance, it touches, like, a lot of the things that we do touch a lot of parts of the business, if not all parts of the business. Like, I mean, design technology is a great example. Is that it? Is that design technology? Is it risk management? Is it practice? You know, if there's a sustainable analysis tool, like, is that. Where does that go? Like, is it under sustainability? There's all these, like, false choices we've had to make. And I think what is nice in the future is I think we don't need to care as much about all that and we can really be more focused on what are the biggest knowledge gaps in our company and how do we fill them. You know, what knowledge does our company not currently have in our heads or written down that we need to acquire to meet our goals? Like, we can focus more on that part of it and worry less about the structure and the retrieval if that. There's a part of my brain that is screaming out that saying, this is a terrible thing to say out loud, you know, because I do think there are still people that browse. Like, it's still helpful. Like, I'm not saying information architecture is dead and nobody should care. Yeah. But it's, it's, it's a spectrum. Right. And it's shifting a little bit.

Well, and there's different ways to do it. And yeah, not. Those aren't necessarily. Like, picking one is not necessarily the right or the wrong answer. There's just a Different way to do it. Like if you go back to the Google example versus Yahoo in the early days, right? Yahoo was a directory. Google just gave you a search box.

Like I wish the people that haven't been on the web could go Google that and try and look at pre Google Yahoo directories. And like this is how people browsed the Internet. It was yes, wild.

Double clicking into folders, right into categories. And then oh yep, not the right one. Go back out, back out, back out a level or two go into a different rabbit hole, right?

And then geocities came out and it's like, okay, so San Francisco now I need to go to the specific GeoCities page for the city that I want to learn about. And now I have to browse that holder.

It also reminds me of just like working in an office and they. And the somebody takes it upon themselves it or some committee to redefine the folder structure for jobs, right. And, and what are the appropriate folders? And, and yeah, like what's the template for that? Where do you put stuff so that people can find it? And, and so I agree, like it's better just to not have to worry about all that stuff. And at the same, because it's interesting because I think about my kids, right, And I think about like they don't even know what files are necessarily, right. It's like they're using their phone. And the stuff that I need is in the app that I made it in or that I grabbed it in, right. And so this, it's interesting like the, the way that, the way that it now works is obscuring that kind of stuff because like it doesn't really matter for a lot of people. It matters for some people, right, but not for everybody. Yeah, right. It's an interesting evolution in the way that information is stored and surfaced.

You bring up a really interesting point there. There's also this kind of like one of the things that's enabled rag or retrieval augmented generation is what's called semantic search. It's oftentimes called vector search. And so when most people think of search, they think of keyword search and the precision required in order to find exactly what you look for. Especially with, I mean Google is a little bit of an extra case because they've been using semantic search all along. We just didn't know it. But if you think about anything you've used inside your business, you have to search on the right term. No typos, you know, in order to find what you're looking for. And if you search on standards, and it's actually a guideline. Good luck. Right. So, like, all that naming, or what we call it. Good luck, it's taxonomy. Right. That's what we call it in knowledge management. So what semantic search does is it understands actually my kind of more shades of meaning and words that are clustered together. And so when we do that process, we have to kind of prepare all the content for retrieval. Augmented generation, essentially what we're doing is we're putting into a semantic search database so that it's easy to retrieve. Um, and that's been. That's made a huge difference. It's way more forgiving in terms of getting precision in terms of exactly what word you search on. Um, and it's allowed people, because if you watch people search, they say all kinds of things, there's all kinds of typos. And so there's this kind of flexibility. Like, how they really want to search is just like, type really fast and then, yeah, if they don't find what, they'll type really fast again. And, you know, just kind of. It's kind of staccato.

It kind of spoils you too, right? It's like, I can't tell you how many times I'm using a tool, like an AI tool, and I. I don't. I. I misspell something and it actually doesn't matter.

Right.

And then you go back to a tool where it's like one letter is off and it's like, I can't find what you're looking for. Like, what do you mean? Right. I'm so spoiled now. In such a short period of time with. With that kind of thing and that. That's pretty. It's different for sure.

It's different. It's different. And understanding acronyms is different. Like, there's so many, like, things that it's unlocking. Like, especially, I mean, one of the audience, one of the kind of core people I think about all the time are like emerging professionals, and they walk in the door on day one, and it is just like an acronym and jargon fiesta for them. Right. It's just.

That's the name of this episode, right?

There it is. It's wild. I mean, we like to do it as people, but I don't know, AEC just takes it to. I mean, even our industry is an acronym. Yes. I mean, that's where we are. So pretty meta. Yeah. So I'm trying to imagine what it's like to be, like, 22 years old and like, so you're in an AEC firm and now you're going to work in a healthcare project. Oh my God, the like jargon soup that's coming at you. So like being able to like quietly under your desk and Google like look up what's a nicu, what's a bxp, what's lead, like without the kind of shame like, and be able to get a good answer, like, I don't know. I feel like that's a really nice thing.

Yeah. Okay, so you mentioned a little bit about clients and the work that you're doing. You have some case studies that you could share about the kinds of work that you're doing and who's doing it with you. I mean obviously this takes kind of the whole AI and the search takes kind of an early adopter mindset. And I'm curious to hear how willing firms were to get into that. But then also who are your champions in this arena?

You know, it's interesting, some of the case studies I've been sharing about what I'm calling KM 3.0, right. The knowledge in the flow of work, AI search stuff, have been doing knowledge management for a long time and they were in a really good position to take advantage of this technology when it came out. So one of the case studies I've been sharing is BWVR. It's an architecture firm in St. Paul, Minnesota. A couple hundred people, multiple, multiple offices. They do a lot of healthcare work and they have built really good communities of practice on synthesis. They call all of our clients, rename their intranet and brand it. They call theirs beehive. So they have built a really strong community of practice. So for this is another KM term, knowledge management term. So communities of practice are getting groups of people together who share a domain. So it could be healthcare architecture, it could be project management, it could be grasshopper, to share knowledge of what they're learning on their projects or their initiatives, to learn faster together. So that's a long worn kind of thing that we've done in knowledge management. They do it, they kind of host theirs on their intranet. And one of the things that they do, because they do, they onboarded 70 people in the last year into the firm. They're emerging professionals and they just need those people to be able to contribute as quickly as possible. And so they created training courses, you know, for those people to learn healthcare. So there's like 18 modules of like the basics of healthcare architecture. It's a lot. They've got best practices, they've got the kind of things you would imagine they share, they have these lunch and learns, et cetera. So they had been doing all that already when AI Search came along. And so it was really easy to go to a firm like BWBR and say, let's take a look at what this is going to do. We think it's going to make that retrieval part that we've been talking about. It's going to completely change the game. And so I think as far as, like, how we got people into beta, we started the first private beta in October of last year, and we started with three firms. And it was like, you know, there's an old saying. I think it was Reid Hoffman, he's the founder of LinkedIn, said, if you're not embarrassed of your version one, you didn't launch it soon enough. Yeah, and I think I'm like, halfway on that boat, you know, Like, I'm like, there's a level of polish that I'm just not going to, like, put something out the door. But there's another thing. It's like, it is. It is wildly done from per. Like, you can see the beginning of it. You see how it's going to be valuable. We're not, like, irresponsible and not caring. But, like, it's not done. And, like, what is important, and I think especially I've learned from working in AI, is we have a really hard time predicting what's going to be important and how long things are going to take. Like, you know, I've been working in software for a couple decades now, and I have a pretty good intuition. Like, we say, hey, we're going to build this thing. I can sit with the engineering team and we can basically say plus or minus this many months, like, or this many weeks. Like, we have a sense of how to build that stuff. With AI, it's wild. Like, you just don't know. And in both directions, like, we've had stuff, like, we're doing stuff in the product today with AI Search that we thought was going to be a couple years out, with us having to build a lot more infrastructure. On the other hand, there's other stuff that's been a lot slower to get right, so it's just really hard to predict. But the other part of that has been we had a lot of hypotheses. Like, we got a 1.0 out and we had a lot of guesses on what, like, 1.1 and 1.2 would be, but we didn't commit to any of it because we wanted to get feedback from real users and understand a what was the most broken. But like, what was the most compelling opportunity for us to work? And nowhere really on the bingo card, I think, did we think that employee search was going to be the next thing we built. We thought there was a lot of other stuff we're gonna have to build first. But because people can rate their searches in the platform and put feedback, we see all the feedback when people like rate how good their AI searches are. And it just became, it was like 45, 50% of our feedback was about people looking for people, which I think is also really interesting. Based on skills.

Based on like, what, what are they? What are they searching for?

Skills licenses. Who do I go to? A lot of it was like, I have a question, especially if you're in a multi office firm, who's the expert? Who's the expert or who's the person who signs off, like, if I have to get a certificate of insurance, like pedestrian, stuff like that. Like, there are hundreds of these questions that come up every day that people need to find out who in the company knows the thing or who's worked on this kind of project before. You know, like all this kind of stuff, like, and what I feel like is really like kind of warming or comforting about that is this idea that like AI is going to reduce kind of the need for people. Like, if you think about it, like we took a swing at what we thought the most important thing to do was this general knowledge base search. And the community told us actually we want it to be able to connect with each other. Like that's the, that's the next most valuable thing that this product can do for us is help us connect with our peers. And then the thing after that is about projects. You know, obviously I want to know how many projects we've done this square footage over the last five years, that kind of thing. But also who worked on that project or who are the people that have worked on projects in the city? Like, it's, that's what people really want to use. We're seeing emerge organically what people really want to use this technology for. So like, I feel like our role is kind of to facilitate the conversation, build the tools and just be really good listeners and follow that puck wherever it's, you know, I know you're supposed to skate to where the puck is going. So we're trying to imagine, you know, that whole thing. Yeah.

I'm curious from your point of view, slight tangent here is like, it seems to me then that like plugging that data in, constantly updating it is super Important otherwise, like, how are people finding the right. I mean, there's turnover, sure. There's retirement, sure. But there's also title changes. There's also different roles. Like a principal could be the project manager on a project.

Yep.

And they have eight other projects that they do a completely different role on.

I'm just curious, like, how does that.

Information work its way in there? Because I think traditionally that kind of stuff is just like, it's just word of mouth. Like, oh, yeah, Ken's doing this thing on this project now and it's not actually codified anywhere. Or maybe it is. I mean, maybe I just don't know where it. Who's doing that and when they're doing it, but maybe it is in there.

Yeah, that's a really, really good, meaty subject.

Too much, maybe.

I mean, it's up to you. This is your show. You tell me in 30 minutes when I'm done answering.

Well, I mean, it's just. I think this is one of the objections. Maybe people, maybe it's in the back of their mind, but it's like, well, when I talk about knowledge management, like, how on it do I have to be so that it's effective?

I think, like, from first principles perspective, like, you can't manage all the knowledge in your organization. Like, you just need to know that that's not a. That's not a reasonable objective. And so a lot of it has to be prioritizing what knowledge you choose to manage based on what's important to your company. So I gave the BWBR example about upskilling folks. But in other cases we see, you know, it might be that they've got some experts that are retiring and, like, they feel like we need to transition. What these folks know about what they work on, like, and we need to start early, not like with three weeks left before.

Yeah.

As they're blowing out the candles. We're trying to, like, transfer knowledge.

See you suckers, right?

See you suckers. Or getting better at onboarding or improving quality. Like, there's all these different things that, like, could be the most important thing for an AEC firm. And the goal is to tie knowledge management to whatever that most important strategic thing is and see how knowledge management can help you move further, faster towards it. Some of it is, you know, this is why integrations are important. You know, a lot of the data that we would want to pull into synthesis already looked in other systems, so. Great. But I think that the trick of it, beyond prioritizing what and doing integrations, is raising the value, we call it return on knowledge a lot. And so when people see that, okay, so if we do capture this, we're going to be able to access it and other people are going to be able to access it, we're going to be able to use it in all these other ways. They can do the math in terms of the ROI and realize it's worth a little bit more effort to get more return, whereas before it was a little bit more suspicious. So I think that's some of it. I think that I mentioned the feedback. What's been good about the feedback. I mentioned the feedback in this AI search, the ratings in terms of how it comes back to us. But the other part of that's important is it goes back to the knowledge managers at our clients. And so they're able to understand, like, instead of constantly being in this state of paranoia that things are out of date and there are. We do recommend kind of proactive. Like, it's almost like a crop rotation scene where you're kind of like rotating through your intranet on a period, three months, six months, whatever it is, and just making sure stuff is still current. Like, there are ways to do that, but the search exposes stuff faster than that. But also it exposes gaps. And so somebody's looking for this thing and nobody. Nobody. We have a client who their values are. I guess every client thinks their values are important, but for them, like, they talk about them all the time. It's like one of those companies, they talk about them all the meetings, they open it with a slide. Nobody had written the values down on the Internet. It is so like part of the way that they talk about the company that nobody actually wrote it down. So then one of the first searches was, what are our values? Sorry, we can't return. There's no information found. And at first they thought the product was broken, but then we kind of like realized it's like, oh my gosh, nobody wrote it down.

So like that kind of exposes those kind of holes.

It exposes the holes and then the knowledge management team can go plug in holes and they can pull and they can go to a subject matter expert with evidence saying, like, look, there's six queries in the last two weeks asking for this thing that, you know, they probably, when they couldn't find it came to you anyway. So, like, what do you think about helping out?

It's like people who create content on the Internet, like the comment sections are gold mines sometimes, right? It's like, like those are suggestions or things that you're, you're not addressing because. But, but when somebody brings it up, it's like, oh, yeah. So it's constantly kind of giving you indicators of what could be next on.

Exactly. We, you know, we think it's a, you know, it's a virtuous cycle or like a flywheel.

Right.

If you improve the content, then you build trust, people will search more, they'll give you more feedback so you can improve your content and they search more like it's this really beautiful cycle. And so when, when it's working. Well, that's what it looks like.

Yeah. It reminds me the parallel now that I'm thinking of is like culture, right. It's like it's the glue of the community. Right. And what you're talking about, like, through this participation of constantly kind of filling these holes, updating, getting rid of the old stuff, you know, a lot of information should have an expiration date on it. So like the cyclical crop rotation thing that you're talking about I think is super, super valuable as far as just like a strategy of how to go through this and actually dedicating people or a team or a person to do that is also like a key thing that people shouldn't not tune into. This is a big job, right?

Can I say something about that? Actually, it is a big job and I think, though, it should be a team sport, not a person's responsibility. I've seen a lot of Internets in my day and I've seen several not successful ones where it's one person's job to upload all the content for the entire company. And yeah, if you've got something new, email it to Evan and he'll put it up for you. And if it's not up there, it's like, well, Evan, welcome this. It's just, it's not. It's not great.

And the byline on everything is this one person.

Yeah, it's all this. One person's byline. You got it. Right. So what, what we advise is definitely getting kind of like, this isn't going to sound like rocket science until the very end. So, for example, if you've got 7 different market sectors or different departments, you want to find a champion with each one of those market sectors or departments. What's important is it's not the. It's not the department head usually, or it's not the practice leader, it's what we call an expert, right. That's the next. It's the person who's going to be. You've Got new jargon. I'm hopefully getting you a bunch. You are. And so think about this for a second. So you've got somebody who's maybe 10 years in. So they're not like an emerging professional, but they're not senior. They're right in the middle, mid career. And so if they're going to help build out the content for we'll just stick with sustainability. They're probably going to interview the subject matter expert, pull information out of their head. They're going to connect and then bitfooters appointment. Just one. So they'll spend time with other experts. They'll kind of own and kind of maintain that material. They'll get to know other knowledge managers or people doing that role within the company. They'll get to know their peers. So they're kind of networking and so they're both kind of developing their knowledge base. They're building a personal brand for themselves as kind of a go to person there.

Yeah.

And it's also good for the expert because they're offloading that responsibility and it's good for the business because they've kept that person engaged, they're developing them, they've done some risk management for when the expert retires or leaves or whatever it is. And it's a sustainable process because when that expert becomes the expert, then you backfill them with the, you know, next man up kind of thing or next woman up kind of thing. So that's the kind of. I learned that from a mentor of mine named Carla o' Dell from APQC acronym. She's like one of the main thought leaders in the space of of knowledge management. And that expert thing is just transformative, I think in the way to think about about doing this. They also have more time like there's all these things.

Sure, yeah, that makes sense. You mentioned one example at least maybe. Do you have any other examples of kind of unlocks that people have seen with these new tools?

Sure, several. I mean I'll use Bora as an example with Corey Squire, who's the director of sustainability. I mean Corey, for many people on this podcast probably know of Corey. He was the ran sustainability at Lake Flato before that. He's been very involved in AIA and defining a bunch of guidelines. He wrote a book in 2023 called People Planet Design which is awesome. So knows his stuff. He saw an opportunity to really make that knowledge accessible inside his company and to kind of like accelerate people's growth and development. To not answer the same question over and over again, but also to free up time for him to do, like, you know, work on more advanced projects, push the edge on sustainability. And so one of the things he did that was just awesome, besides building out best practices and sharing knowledge and videos, is he uploaded his book to their intranet, which was pretty awesome, just chapter by chapter as a, you know, a document library. And now that's all searchable to the company. So now if Corey's at a conference, if he's traveling, he's out of the office, people have access to what we affectionately call RoboCory, or kind of like his digital twin that he's built. And other experts at Bora as well. And so this idea that, you know, I've been, I've noticed since we put AI Search on our version of synthesis, written probably more in the last year on our intranet than I have in the previous five combined, because it's that same thing. It's like I'm getting confidence that I know that this is going to make its way into people's hands. You have some incentive, I have some incentive. But also, and this is the part that I think is really important is it's so much easier. Like, I use AI a lot to help me transfer my own knowledge, right? So I use it in design all the time to help solve design problems. Like, for example, we're working on a learning management system to add into synthesis. And there's a lot of thorny design problems in a learning management system. Like even get into assignments and how they work and what happens if somebody drops the course. Like, there's all these things. And so I end up workshopping a lot of that in voice mode with ChatGPT. Back and forth, you know, back and forth, back and forth. And what I always do at the end of one of those, like once I've kind of gotten to a defined design direction, is I have it summarize the different avenues we explored, like why we made the decisions we did. If we decide not to enable something in version one. What are some known workarounds? Like, I basically do a debrief of the design session with ChatGPT. Yeah, right. And then I add that to our intranet. So, like, that context, it's not just what the decision was and how it works, it's why who like all that stuff. And like, yeah, people want context, robots want context. Like, more context is good. And so I think, like, the, the.

AEC version of this, to me as a designer, is like, these are all the layers of trace that led to us actually drawing it in bim.

Right.

So yeah, it's like a lot of times that stuff doesn't get scanned or just gets tossed away and then you don't know why, you know, or it was in the meeting minutes because it wasn't an official meeting. Right. And it's like all of these reasons why this is a really cool example of how that can actually become a superpower later. It's like, oh, okay, so this is the way it is. But this is why it's like that.

This is why it's like that.

That's, that's amazing. Yeah.

Think about like I mentioned Corey, like I might not get this right, but like, let's just imagine like they, they defined a window wall ratio optimal of being 30 to 50% or something like that. The comments can let me know how off I am.

Yeah, 80%. Yeah.

Yeah.

Right.

But if that's, if it's just that fact, and I'm not saying this is what happened on Boars Internet, I'm just using it as an example. Like I don't know why it's 30 to 50%. 30 to 50% is a 20% range. When should it be more like 30? When should it be more like 50? Like what do I. Does it different in different conditions and different. Like there's just so much more context versus like this is just the rule.

Apply it. Right.

And I think the more we can capture that, that helps get better answers in AI search, but it also helps our employees actually learn and understand instead of memorizing some facts or just blindly following a rule. So yeah, I think it's going to, I think KM 3.0 or AI is going to pull better knowledge sharing out of us. Like it's going to, it's going to ask that of us to be better at sharing context and better at staying on top of our knowledge in order to take advantage of. It's a, it's a trade. Right. In order to get the advantages of these new tools, we're going to have to do some of these things. And I think they're all generally net positives.

I mean that the thing that you just did seems like kind of a gold nugget. The example that you just shared about your process. What other kinds of things are firms going to have to do to actually have this? Be. Have the most impact. Be the most useful in their firm.

Yeah. Their strategy, I think. Well, I mean to state the obvious, I think actually this is an interesting. I might take you down a little. We might meander a little bit on this answer Where I was going to say, and I will still say is that high quality firmwide data is like really important. And, and I say high quality and firm wide because at least in our case we'll go back to silos. We've kind of, since we've been in business and honestly since before I was even in knowledge architecture, I've kind of really oversimplified the world into kind of project data and like kind of enterprise or firmwide level data. So. Meaning the standard versus what you're doing on this project at sfo. Right on, you know, specific. And so, and generally the way that we've architected systems has kept those worlds relatively separate from each other. And so if I'm searching on a tiling standard, if I do it on synthesis, the goal would be I want to find like what we think about the best way to do tiles is or do indoor air quality or window wall ratio or whatever it is. If I'm searching project content like it would be probably something more like where are all the projects where we've used this particular tile? Right. Or what are all the projects? And it feels like there's a lot of. We've gotten some like a little bit of pressure from our community to bring those two worlds together and have one search bar to rule them all. I don't know that that's right. Personally, I think. And so there is, but. But it does require somebody somewhere to decide this is what good looks like for our company. Like this is, you know, you can get into learning and development, this is the way we do project management. Or these are the three ways we do, you know, like there's some variants, maybe it doesn't have to sound so totalitarian as I made it sound, but to really kind of say like this is how we do things here, that's a lot about culture and that's a lot about knowledge management. So I think taking the time to communicate and capture how your firm works and why it does the things that it does, like that doesn't. That's. That's effort. So I think it's going to ask us to do that. I think it's going to. I mentioned the feedback cycle before, like that where you know, you get new feedback, you fix the knowledge, people get better search. That only works if you actually make changes on two levels. One, the content doesn't get it better if you don't updated. But also I think the feedback slows down too. Like I've always thought this with like surveys and feedback or asking for input from Anybody, like, if you don't show somebody that you've taken action based on the feedback that they took time out of their day to give you, like, guess what I mean, that feedback's going to start drying up. Like, you know, and so I think there's, there's that kind of responsibility to follow through and like, keep that continuous improvement machine going.

So what do you think that actually takes for a firm then? I mean, if you were to kind of just lay out, I mean, obviously it depends is like the answer, right? But if you were just to kind of. Let's just pick a scenario. Like, for example, friend of the show, Adam Wilbrecht was the first chief Knowledge Officer that I had ever heard of. Right. And I'm sure he wasn't the literal first, but in aec, he was the first to me, and when he was at a previous firm and it was like, what does that mean? And so I think part of it that you're talking about is like, people who do a thing. There's like a team kind of, but there's also kind of a structure and representatives within different departments in a firm and all these things. But I mean, if you could legitimately just kind of lay out a basic strategy to set an expectation. Because this doesn't happen by itself, right? And garbage in, garbage out. You don't update, you don't, you don't improve based on feedback. Like, these are all indicators. But I think it still takes, like, people to actually, you know, you got to say, this person does this. How much of their time do they spend do it? Like, what have you seen as far as kind of like minimum viable product from a staffing standpoint to make this work in, in your favor?

Yeah, I mean, I think I like that you ended on minimal vibe. So we have an asset we developed several years ago called the Essential Content Checklist. And that's kind of what it sounds like. Like, I think we developed it when we had like 100 clients and we basically looked across those firms that were successful and we tried to just kind of just generally get like, what are the kind of key pieces that every AEC firm should have on their intranet? And it, I mean, it gets done into very basic stuff. Like, when do I get paid? What are our holidays? You know, what's our technology, how to like all that stuff that's just like, that's just kind of like a floor. Like, we're like, just do that. Because the second you turn one of these systems on, like, that's the stuff people are Going to reliably always search for and find and need, need to get access to. Otherwise your HR department and your marketing people and your IT people and operational, like they're just going to get inundated with the same, same kind of questions. So like that's really low hanging fruit at most firms. They have it somewhere, you know, and so this, this checklist ends up becoming kind of a treasure hunt, you know, like maybe some of it lives off in this other old Internet, maybe some of it's on the network share, you know, maybe some of it's all just sent via email. And so a lot of it is kind of consolidating that, updating what's missing and kind of bringing it up to like good level of quality and then continuing to maintain it going forward once you get that in play. And then we generally recommend that firms just get really good basic employee data together so you know who does what, who, you know, who works in what office. Like really simple, like again, simple stuff. And the reason that's important is it's not that hard to do, but it builds trust in your community that your data is good and that's what you want to start doing pretty quickly, is establishing trust. Then it starts getting into some pretty interesting territory. So I talked about knowledge sharing before maybe, and this is where the it depends happens. Maybe what the most important thing to do at your company is, is like you have a new strategic plan, your leadership is trying to steer the culture in a new direction. There's just something kind of big happening at the company and you want to use knowledge management as a way to drive that message through the organization. So everyone's aligned, they know what the important things are. Maybe it's like you have a really robust design culture or you want to have more of a robust design culture and you really want these like people sharing stuff from the model shop, people sharing stuff from project tours or site visits or talking about crits. Like so then you start kind of like picking on like, who do you want your company to be in a year from now or two years or three years from now? And you work backwards. So like what are the knowledge management interventions to make that happen? I mentioned the upskilling thing from B2B. So like we need to get these health picker people up to speed. So it does vary very quickly. And because the goals vary, then the resourcing and the staffing vary quite a bit too. So if what you're trying to do is get a design conversation going, you can imagine that's A very different set of resources than trying to improve quality or trying to upskill employees. So you usually have an executive sponsor. So they kind of sit at the board or management team level. They're the connection to the knowledge management team of like what matters and what doesn't. So they can keep the knowledge management team from chasing like science projects or things that sound interesting. They can say like, this is the important thing I would spend time here. You usually have a champion or two who kind of like orchestrate the work of a bunch of other people. And then depending on what the initiative is, you could have five. We're talking part. No, this isn't like generally full time roles. Like this is like part of somebody's job. I think the more, I think you would need a good clear structure, but I think the more people you have involved in general is better than having like, if you could have just like a couple FTEs doing this all the time. Like I actually don't think that's great. I think it's better to have this be part, you know, piece and parts of people's role with a good sponsor and a good champion or two, kind of just overseeing the machinery to make sure that it stays on strategy and that it's like operating efficiently, if that makes sense.

And in your position, I mean, as you built this platform and you have lots and lots and lots of firms using it, it's fantastic to have you as a resource to say here's some best practices for implementing this so that it gives you the return faster than you trying to figure it out on your own kind of a thing. So that's, that's a, that's a big plus.

Well, yeah, and I think, I mean, I can't overstate the, you know, one of the things I love about our industry that confounds my friends and even my wife who's outside of it is how open we are to sharing with each other, even with competitors. Like it's always like I'll go to some a non AEC thing and talk about what we're doing and like how in the world are you getting these companies to share this kind of stuff on your webinars and your conferences? Like they figured something really cool out and the first thing they do is they turn around and tell the rest of their community like what are they doing? And I get it though, actually. I think like, you know, we're not competing on the way that we do the knowledge sharing piece or this piece of this grasshopper script that we've written. Like we're competing on something else. And I think we're moving further, faster as an industry together by sharing. But what that's allowed Knowledge Architecture to do is like, again, like, I feel like we're still kind of just hosting this conversation with people who have pieces of the answer figured out. And our job is to keep that knowledge flowing and to synthesize it and be able to turn it around. But, like, we don't, you know, like, we can't give those recommendations if it weren't for how generous and transparent our community is. So it's, it's an amazing, amazing thing. We don't take it for granted.

Can you talk about just kind of the. Is this a heavy lift for people? I think when people think about, okay, now I've got to put the data in the right buckets, like severance, right? You click on the number and it goes into the. Nobody knows what it means. But it's like, there's this, this fear of like, data entry, like, menial work kind of. Okay, I got to make sure. But I think it's probably not necessarily like that. At least. I'm curious of your perspective from like a structured versus unstructured. Like. Like, do we just throw the PDFs of the manuals in there and AI can just pull out the relevant information, or do we need to be more structured about it? What if we haven't been keeping records very structured versus, you know, like some. Sure, some firms do it very well and there's a whole spectrum there. Right. So when, when somebody's looking at implementing something like this, like, what kind of a lift is it for them to start having something usable to pull from?

Do I have to say it depends for you or like a man, it is. It's the answer to almost every question that comes up like this. But like, to put. To put some more thought behind that, I mean, one of the, One of the variables I would look at for it depends is your kind of existing state. So we've had clients come on board who had a very like, maybe in a more analog way or a less efficient way, organized, really good resources, and they just needed a really good bucket that was searchable and easy to use to put them in. And that's been a really easy transition. And they came in and their Deltec data was great, and their open asset data was great. And we just basically plugged it all together and off they go. Not quite that simple, but that's one end of the spectrum. We've had other clients who have come on Board who? Like, I'm thinking of one. I'm not going to name them. I'll let them tell. Maybe they'll tell their story on this. But like, I'll use them as an example. They came on board at like 60 people and had nothing like in terms of systems. And they had just, they had, they had severe growing pains and wanted to become like they were getting all this work, they were growing. They're like, we cannot keep operating like this without really clear procedures and standards and a way to connect the people. Like it's just not working anymore. And so they basically, they took a year to implement because they had to write everything. Like these are things that they had never, ever written down. It was all just kind of social or tribal knowledge. And so that was actually the lift. We've had clients use synthesis as kind of like a Trojan horse for the culture change they were wanting to effect. So I can think of one CEO who took over one of our clients and it's an engineering firm and they had kind of a. You know, I say this, I'm going to put a disclaimer out here. Some of the most collaborative, sharing knowledge. Best clients we have are engineers. Like they, they do really, really well. Like with our product. This culture was kind of more of the caricature that like, I don't know, some architects may say about engineers. Right. They just were very close to the vest. They weren't sharing like, it was a tough environment. And she came in and basically, like, collaboration is the key to our thriving and actually maybe even survival as a company. And so she used this kind of social intranet at the time as a way like she kind of like used sharing on the platform and asking all of the kind of leaders and held them accountable to sharing X amount of time as a way to kind of make that culture change of making the company more collaborative and open and transparent. So I'm kind of a little all over the map there, but that's honestly what our journey's been like. It's like we just find people in various states of existing content and also what their ambitions are and then how much they can resource the thing. So sometimes people try and take this on with like three other things going. And some people wait until they can really get a strong Runway at it.

And so when it, when it comes to like implementation.

Yeah.

How much of a resource are you available to them to help figure that out? So it's not like buying a piece of box software off the shelf. You're on your own.

Yeah, it it matter. It depends how much they want us to be. So like we can run the gamut from like we do strategy, we help you figure all this out, we help you with best practices, we like help you with the information architecture. Like we can drive really, really far for somebody. We've had clients come on and like, nope, you guys have lots of good online content, you've got lots of good training videos. Like it'd be good to ask you questions when they come up but they prefer to be self directed and so we can support people at that kind of high touch level or kind of a more self directed level. And it's just interesting. And like some people I mentioned our community beyond the conference. We have like quarterly roundtables. We get people together to share and we have monthly webinars. There's like, we have an online community where people can ask questions and there are clients that don't take advantage of it at all. But their Internets are really healthy. And so okay, you know, like we're just going to keep building great resources and it's really a matter of like where. And maybe they were really into it at one point and they drift out or maybe they didn't start into it and then later they're like we need to do this thing and I can't do this alone. So. Yeah, yeah, it's varies for sure.

Interesting. Well, I'm curious about. I guess we could start to wrap up here. But what, where's this going? Like what do you. Things have changed so quickly in a short period of time. I remember you showing like the periodic table of all of these different pieces of knowledge management inside of firms and something that you brought it up in this tangentially you weren't directly relating it back to that diagram, but it's like you can't do all of these things. But I think what's interesting is we're starting to see for the people who do a lot of them, maybe they're not doing a lot of them really well. This is starting to connect the dots between a lot of those because of the evolution in a short period of time that we've seen with AI. But where do you see this going? I mean crystal ball time maybe a little bit or maybe there's some stuff that's more rooted in reality that you're seeing that you're really excited about.

Yeah, I mean in the very short term we, I mentioned we. The feedback we got about employees. So we built that kind of advanced employee search. We're now building the advanced project Search as well. So that's, that's exciting. We're building a learning management system into synthesis, as I kind of touched on before. That's going to obviously have a big interplay with search. And I mentioned, I said high quality, firm, wide content. Well, there's a lot potentially in a learning management system. So to have the intranet, the LMS and the AI search all under one roof, like, we're pretty excited, you know, about where that goes. So I'll answer your question maybe in two ways. On one level, we don't plan out any further than that anymore. I used to do that more like when I first got started, I used to look out kind of like three, four releases and it was fun to kind of roadmap all that out. I have learned over my career that, like, that's just not great product management because, well, two things can happen. One is, let's say you make a mistake. Let's say that you think you're clever and you know what you're doing and you tell people about it, and then you get to the moment and it's no longer the most important thing to do. So you either break your promise to your community or you go ahead and build the wrong thing because you don't want to break your promise. So I don't know, like, other than kind of bringing the LMS into our search platform, like, I've got a long list from client feedback and I can guess, but like, we don't commit to it and I don't, I don't know. I do think it's going to get to be. I think we've gotten to this level of quality pretty quickly, but I feel like the expectations of users of what AI is making possible is just going to keep ratcheting up. So what I know is we're onto something with search and discovery. I think that we're going to be working on it for years and it's just going to keep getting more and more interesting based on our ability to kind of scaffold out new technology. But also the underlying technology just keeps unlocking these new things. Like, for example, two weeks ago, ChatGPT 4.1 came out and didn't sound that exciting. 4.1, how revolutionary could that be? It is way faster and it's really, really good following instructions. And when I had mentioned that agentic rag stuff before, like our engineering team, especially our director of engineering, like, a lot of the way that these systems get built is like you're giving context to the model to tell it how to work with the data that you're giving it. And like, sometimes it's like talking to like a five year old, you're like, no, seriously, please do not do this. You know, like when this thing, like he's shown me some of these prompts. Yes. Like, it's literally like your interface to the model is just context like that. And so you have to get really clever. And so this 4.1 model is just like much better at following instructions. So we can dial back some of the all caps, you know, and the unnecessary brackets. The brackets, right, exactly. The exclamation points. And so it's like, that's a little thing. But like these models, like one of the things that we're very excited to do is build an aec. I've mentioned semantic search before, so without going too far into a tangent, that's built on a technology called embeddings, which basically take every piece of content and they kind of like, if you imagine like DNA, they kind of embed it with a unique code that basically means like sustainability ish or contracts ish kind of thing. Well, we think there's a huge opportunity to build an AEC specific embeddings model that will understand our domain and our acronym and our jargons and have a much better semantic understanding. Like, and it's AEC specific, but also kind of professional services specific. Like the generic models that do semantic search, like, they don't really understand that utilization and chargeability and these kind of words, they don't really know what those mean. They were kind of trained on more consumer data. And so but right now, like, it seems like that's not really commercially viable. Like there are tools out there to do that, like in the lab, but it doesn't feel like it's prime time stuff yet. So we're kind of monitoring that as an opportunity for when those tools become like repeatable and actually something we'd want to bring into our product, like, we would jump on it. So some of it's just like waiting, you know, like carefully monitoring the evolving tech landscape and looking for matches between the things that are most important to our clients and the emerging technology and be able to like kind of do that matchmaking. So the thing I know we'll be doing is staying very close to the feedback into our community because they generally lead us, you know, in the right direction, or at least we walk in the right direction together.

Yeah, exciting. Well, what I'm also excited about, Chris, is just kind of this, we're going to do a Multi part series with this that we're sharing. So this is the first piece and this kind of kicks us off and sets the foundation for. Yeah, we're going to be sharing some stories with the firms that you've been working with and talking about implementation, return on investment, things that have been exciting to them. What are they getting out of it? Was it worth it? Like there's so many questions I think that are going to come up from different perspectives and I think that's going to be really. I mean I appreciate you guys investing into telling this story over many different episodes because there's a lot of nuance. But what I'm most excited about is that we're not stopping with just kind of this, what feels like of, you know, we haven't gotten to the really deep implementation parts and the boots on the ground kind of side of it. So for the listeners out there who are wanting to hear more about making this a reality, you're going to hear some real stories about that of people who have been through it and done it. And do you want to kind of talk about the kinds of firms that we're going to be talking with?

Yeah, I mean, I think, I think at a broad level I want to also make clear that I think our intention is to explore knowledge management in general, not just kind of their implementation of synthesis. So I think it's important that we kind of fit into an overall ecosystem or a knowledge architecture, if you will, of like how they think about knowledge management. I can speak to three of the guests that we're going to bring on that we know of and there's a couple more that I think we're going to look at down the road. We're getting Ellen Bensky from Turner Fleischer to talk about it and she's their CEO and we'll kind of give the CEO perspective on knowledge management and why they've invested not just in synthesis but in a. They built a knowledge management department and a team and as well as a learning and development team and kind of like why they did that at a 250 person firm. We're going to have Katie Robinson from LS3P who's their chief marketing officer and leads their knowledge management program. Katie has a ton of great, really kind of like detailed like a lot of the stuff you were asking. It's like how are we actually going to get good project data on a sustainable, in a sustainable way? Like Katie's done some smart stuff there. She's done smart stuff with, you know, kind of critical knowledge transfer and experts sharing and a bunch of other things. So that'll be fun to talk to Katie, kind of from that marketing lens on knowledge management. And then Corey, who I'd mentioned from Bora, is going to talk about that sustainability slash subject matter expert, sleep slash practitioner, you know, kind of impact on knowledge management. Different firm sizes, different scales, different positions. But you can see each of them though has led knowledge management in their organization. So it can happen from all different parts of the company depending on your unique business. So I'm excited. These are good friends of mine and good friends of our companies. They've all spoken at our conference K Connect over the years. So I'm excited to bring this to your audience. I think it'll be they'll learn a lot.

Ellen's been on the podcast before. If that name is tingling anybody's ears out there. She was talking about knowledge architecture and that's when I think I became actually acutely aware of that this was a thing in architecture firms. Was going to a class that she co taught with one of her principals from Turner Fleischer at Autodesk University many years ago. And so impressive. Like the level of thinking that I am assuming all of these guests are kind of bringing to the table here is just going to be worth the listen. So really looking forward to this series. And Chris, we'll put links to where people can find out more and connect with you online in the show. Notes for this episode and thanks for coming on. It's been fantastic.

I was glad to be here. Thanks Evan.