INTRANET TOUR | AI-Driven Knowledge Management @ Boulder Associates

Hello, welcome. My name is Susan Strom and I'm the Chief Client Officer here at Knowledge Architecture. With me today is Todd Henderson, Director of Practice Improvement at Boulder Associates, and Christopher Parsons, the founder and CEO of Knowledge Architecture. You are joining us for a KA Connect Deep dive where we share stories of how clients are using synthesis to help evolve and grow their beliefs through knowledge management. Today we're going to get like an AI smorgasbord from Todd. We're going to hear how he uses AI to help translate tacit knowledge that is in experts heads into explicit knowledge assets that other people can use. And then we're going to hear how he's using synthesis AI search to mine that knowledge and get answers to the people who need them. To get us started, I think it would be good for us to clarify what synthesis AI search is and when it will be available. So, Chris, I will hand it over to you for that.

Yes, for those of you that are clients of ours, don't be confused, Todd is going to be showing you things you do not have access to yet. Todd and Boulder Associates are in the private beta for AI search, which is coming to public beta, meaning anyone that wants access can get it in December. If you aren't a Knowledge Architecture client, it's kind of the same thing. You're seeing an early beta of something that will be available in synthesis starting in December, something we've been building on for the majority of the year. And we're very excited to be sharing with everybody. Todd, I think that's it. I think I'm going to stop there and just pass it to you. Show us what you've been up to.

Lovely. Thanks you guys. Thanks everyone who's attending. So super excited to be sharing what I'm up to. So this is me. I've been with Boulder associates, gosh, oh, geez, 22 years this month, actually. So for a minute. And I've held kind of every role, I mean, been here for a significant chunk of my career and so you'll see some of that come through in a little bit. But my current role, I've stepped away from being an office director and, you know, kind of a practicing architect. And I'm now director of Practice Improvement. So I do things like this very briefly. Boulder Associates, we are an architecture and interior design firm nationwide. Practice about, gosh, 240 people today. We work in several market sectors just to give a little context to kind of scale and what we're up to. And we are proud, proud subscribers to Synthesis. We've worked with knowledge architecture for a long time. I think we are actually their first deployment of their synthesis intranet product. And we're very proud of what we've built. We call it Biscuit Mint. And so I'll introduce biscuitnet to you briefly. Here's what it looks like as of, you know, probably yesterday. And let me show you what the AI search does. So maybe to explain, like you might be thinking, biscuits, why biscuits? What's this whole biscuit thing? We name everything after biscuits at Boulder Associates. And so searching with synthesis AI search now gives a really wonderful answer to the question, please tell me about biscuits. What you'll notice right off the bat. Well, you may notice at the end of each paragraph there's a little number in a bracket and if you hover over that, it's a link, it's a live link to content on some, some spot around our intranet, around our site. And so search has, has given this kind of 360 view of this concept of biscuits that sourced from six different locations. And the truth is, you know, as someone who's been at the firm, you know, for a long time, I can read this and go, wow, that's right. And that's also kind of. That's not an answer. That answer, you see, is not an answer that you could just find on some specific page. It's a composite of everything. And I think that starts to point to what is, to me really exciting about, about this advance in, in how our people are going to be able to relate to the knowledge on our, on our site. Let me demonstrate a more, you know, sophisticated, more, a more substantial thing. Maybe an architect is just getting into understanding contracts and they're like, tell me about liability. Pretty complex topic here. I'm hovering over the, the various sources. It's sourcing from posts, it's sourcing from pages. We did a University of Boulder Associates presentation earlier this year about key contact provisions with our insurance carrier. I love this. I love that it, that it actually gives again, you know, that second paragraph is talking about liability in a completely different sense than probably most architects are thinking. We're thinking about getting sued by a client. This is, hey, with AI stuff coming out, we need to be careful about kind of our AI policy on the work that we're giving. So we're not doing, you know, you know, copyright infringement or things like that. And this, so this, this, this technology makes it easier for the person to tunnel into whatever they were interested in or just maybe this is what they need. Okay, so that's sort of the intro to me, the firm and kind of the, the our intranet, I'm assume it's way out low earth orbit and I want to talk at a high level about systems. Systems thinking is a, is a thing I've come across that has, that really informs a lot of how I look at organizational change and understanding how to, how to get from where we are to where we want to be. W. Edward Deming has This wonderful quote. 94%. It's a very specific number. I don't know why it's 95, but 94% of problems in business are systems driven. How much of our management of what we consider to be a problem in our business do we ascribe to the system of our business and not to some person who is not doing the right thing? I think we tend to have, have it inverted. So I think Deming, Deming gets it right. So let's talk about systems. So the things that we tend to manage are the things that we tend to see events, right? Maybe we're unhappy one day because people aren't filling out their timesheets on time. So that's an event that we see. That's a, that's a, that's what's poking up above the water. But the key insight of systems thinking is that those events are, are truly the tip of the iceberg. There's all this stuff below the water that is actually incredibly powerful in informing what those events are. And so if you're just trying to push on the events to make them be different, you're pushing against something that ain't gonna move. Another insight of systems thinking is that when you push on a system, it pushes back. Systems regulate themselves so very often. Well, I'll come back to that thought in a second. Systems thinking has these diagrams oftentimes that people will draw called causal loop diagrams. And here's one about systems. It's a little bit meta and it's just that actions lead to results which lead to actions, right? There are these loops, these feedback loops everywhere they're all over, you can learn to spot them. There's a pithy saying that I think makes a lot of sense, which is today's problems are the result of yesterday's solutions, right? That kind of gets at that, you know, chicken and egg nature of the way systems work. And it informs and it changes our thinking, right? About looking for patterns and structures and not just yelling at someone who didn't fill out their timesheet. Danila Meadows was really, really cool, pioneering systems thinker and she wrote this wonderful paper. It's available, you can Google it, you'll find it for free online. It's actually a really nice entry point into systems thinking. It's called leverage points, places to intervene in a system. And so she uses this metaphor of, you know, like I talked about, we tend to, we see events, you know, someone didn't fill out their time sheet and we want to like yell at them. And that's, that's at the low end of this kind of lever in this metaphor. You want to. Oftentimes that's, that's not enough to get a system to change. So there are. She talks about different places you might intervene and that they have different, a different amount of leverage to, you know, to retune a system to behave differently. So let me give, here's, here's the full, here's her full kind of model here. Not going to go over this in detail, but I want to pick out three things and I'll start closest to the fulcrum, the black bold one. Stocks and flows. That's systems jargon, but for our purposes you can call that like events, things that happen, someone filling out a timesheet, something like that. Midway up and in red is information flows. And I pick that out because I'm gonna, I'll be. That's what I'm kind of talking about today. And I think as knowledge management folks, this is where we put a lot of our thought. And in Meadows model, I mean that's a, that's a. The creation or improvement of a flow of information is a pretty strong leverage point to get a system to do something different. Sometimes giving people access to a stream of knowledge changes a whole lot of stuff. It's kind of cool. Maybe some of you have seen that in action. The last thing I'll pick out is way up towards the top mental models and there's an important thing there. And I will kind of come back to that. But the beliefs that we hold as individuals have a lot to do with the behaviors that we are going to exhibit in the world. The beliefs and values that our organizations espouse and actually live are determinant to everything else that's going to happen. So sometimes that's the place you have to get leverage to make things move. Let me give a more concrete example. Hopefully this is something that maybe people will sort of have some, have some lived experience with. I, I know I do. Managing a healthy weight, right? If you have ever struggled with your weight, I certainly have. Then you have been told, well, all you have to do is eat fewer calories than you burn. You're like, huh? And yet somehow that isn't enough of a leverage point, is it? At least not. It wasn't for me. Not for most people. That's right down there. That's just sort of saying, eat less. You know, put the donut down. It's not enough. Often times. Well, what points of leverage might be stronger in this model that Meadows sets out? Here's some, right? And they use the same color. So observing your weight versus your caloric intake, that's an information flow that can be a very powerful motivator. And in fact, the next one up, frequent weigh weighins and food logging. There's a, there's an author, I forget his name. He wrote a book in the last year or two where he tried every diet he could find, including like some really stupid, I mean, overtly stupid diets like eat nothing but cabbage or that kind of thing. And his, his concluding takeaway is, look, you have a diet and it requires you to weigh yourself frequently and keep track of what you eat. Probably going to work. That is really where it's at because it's creating a data set that creates an information flow from a systems point of view, powerful, powerful leverage. Last point on this example, just to kind of drive the point home, is that's all well and good, but, you know, if a person has limiting beliefs about weight loss, about weight, about their value, about whatever could be a thousand things, then nothing else you do may be enough of a leverage point to, to change how that system, that human beings as a system works. I hope that was a kind of a relevant example to some folks of a way to think about this model. Let's turn it around and kind of come back to the theme of today where probably a lot of us feel this way. Our subject matter experts aren't sharing their knowledge enough, Right? Dang those guys. So sometimes the instinctual management responses yell at the subject matter experts. You need to share more. So that's, that's an example of trying to manage kind of at the event level, at the individual behavior level. What I would ask folks to do is suggest what are some interventions that maybe lie farther up the lever here from a system's point of view, that might be more resourceful, might offer more leverage, more insight into the situation. So I'll stop talking for a minute. I've got an eye on the chat and I'd be curious if people have any ideas.

Todd, could you restate the question one more time?

Yeah. What so if, if what we want to do in an organization is get subject matter experts to share their expertise more. Right. Get it spread around, not locked in their heads. And we acknowledge that just yelling at them, grumbling at them to share more is an insufficient point of leverage. What might we do that is higher up in this kind of. In Donella Meadows model pizza has been suggested. Bill. Welcome. Bill. As a leverage point. Sure. Pizza, maybe Meadows missed. Missed pizza as a, as a point of leverage. I like that. But incentives, right? Certainly. So I'm seeing cool stuff. Show how they're sharing benefits them in the firm. Yeah, that I would, you know, I would sort of call that kind of an information flow. Have anyone else noticed that when you ask people to do any kind of data entry and they can't tell where it goes, it just from their perspective, it goes into a hole, they will stop doing it. But when they see, wow, this actually makes a difference, that's a powerful motivator. Lead by example and share yourself first. I love that appeal to their interests. This is great. This is good stuff. So here's. Whoops, gotta get back to the right screen. Here's a few, here's a few that I, you know, seated here. And I'll start at the top. If your organization doesn't value learning, it may not matter. Anything else you do may not matter. Because the system, if it doesn't value learning, it does value other things. And everything else about the system is going to be responding to what is a value. Not just an espoused value on your website, but like an actual lived it value. You could have a goal as an organization, perhaps from a strategic plan. Holy buckets. We got a whole bunch of people about to retire. Ah, we need to get on the move to train some folks up, remove barriers. Right. The structure of the system. Why is it so hard for subject matter experts to share their knowledge? Maybe, maybe there's a structural reason. And spoiler alert, that's what I'm talking about today. I'll give another one. You could have a rule. Rules in a system are powerful, strong, a strong leverage point. So you could say, hey, you know what, that's it. You don't get your bonus unless you share 12 times a year. Anyone see a problem with that? So if we say that today's problems are a result of yesterday's solutions, what are tomorrow's problems? You want to be careful, right. Intervening in systems because oftentimes we push the wrong thing or we push a thing the wrong direction. If you make a rule like this, you will get that thing. People will contort their behavior like, okay, if that's what's important around here, I'm going to share something and you're probably going to regret that choice. So just a, a note of caution. Okay, enough theory. We're, we're zooming in. I want to talk a little bit more practically about kind of this topic, right? We have experts who know a whole bunch of great stuff. We want that spread into other people's brains. How might we go about that? So, you know, I've been studying experts and I've compiled this list of facts about them. I think it's pretty good. I think it kind of covers the territory. I've noticed experts love talking about their expertise. As an expert on some things myself, I, I love it, love it, love it, love it. It feels good to talk about what you know about and to share it with the world. It's just pleasurable. It's not work. It's joyful. That's me. But I think that's not uncommon, right? Man, is it hard to get an expert to write about something, however, unless they happen to be an expert in writing. But that's kind of. If you went into architecture, it's a pretty safe bet you're not good at writing. So if you buy this, this model of true facts about experts, then it implies that. In the structure of things, maybe the act of needing experts to write stuff down is the barrier that is limiting how much knowledge transfer we're actually able to get. And surely there are other ways to do this, right? All kinds of, of discussions about transfer of critical knowledge and shadowing and things like that. I'm sort of, I'm sort of today focused on kind of capturing stuff and getting it on the Internet. So by necessity, it's, it sort of winds up being a lot of writing. So what's the problem with writing? Just to state the obvious, it takes time and focus. You can't multitask. You, you, you gotta hunker in and sort of do it to do it well. Another true fact about experts is they don't have a lot of time and focus. Generally, those are both in short supply. Being skillful at being a medical planner or an architect or whatever is not the same. Does not, does not imply skill at writing. We get skillful at the things that we do, we get skillful at the things we spend time doing. Here's another sort of system insight, right? So you're good at being an architect, so you get more practice being an architect. So you get better at being an architect, so you get more practice being an architect. Never in that loop are you getting better at writing. And finally, the tyranny of the urgent billable work, clients, deadlines, that stuff is urgent. It's not necessarily important from a strategic point of view, but it's always going to swamp stuff that doesn't have a fixed deadline. So how do we get around this? Stop having experts write stuff. Right. So, and this is not a new idea. And you know, sort of before, you know, this year, last year, whatever, you know, you would hear about firms who would have ghostwriters, they'd have a writer on staff or just someone who was a competent writer or just had more time to write, interview somebody and then do the writing for them. It's expensive. We've, we, you know, it's expensive. We can do better. You know where this is going. You already know where this is going. There was a, a talk given yesterday by APQC that I attended and they shared this really interesting graph about how far AI has come and how quickly in a whole bunch of mostly writing related tasks, reading, writing comprehension tasks. And they say that like these models have, have met or exceeded human competence at these things. I think some of us are probably, nah, it's not my, that's not my experience playing with them. But I think the, the larger point is, probably holds true that this is a good time to be using software to write stuff. So how do we do that? What I'm showing here is basically the, the first step is get an expert talking. They like to talk, they're happy to share. That's a low lift. Get it on video, transcribe the video, summarize the transcription, bada bing, bada boom, off you go. So I'll show how I do that. And just to make the point, once again, these AI tools make some of these steps kind of nearly effortless, frankly. So zooming in further, let's talk tactics here. And first I want to introduce you to Darcy. This is Darcy. I hope all of your organizations have a Darcy because our Darcy is amazing. She is my partner. She started Orange county office. She is a absolute force of nature and work. A wizard as medical planner, healthcare architect, strategic thinker. We all need. Darcy's right. She's also incredibly busy. Right? As a result, she is incredibly busy. And so I had, I, I called Darcy, oh two, three weeks ago because I got a, got a weird email from a client that we, that we both do some work for. And so I was calling to kind of ask her like, what's. What's going on behind this? And long story short, there's like, this kind of political thing going on between a person who works with this client, working on legislation that was kind of trying to thwart the big state agency that manages healthcare projects in California. And it kind of. The wheels were falling off the wagon and things were getting weird. So she's telling me about this, and then she sort of pauses, and she has this deep sigh like, I really need to share this. I just. When. When am I going to find the time to share this? And so I was like, you know, you just talk to me about it. Let's. Let's meet. I have a process because I've been kind of developing this. This workflow on our corporate university sessions and, you know, kind of other opportunistic, you know, moments. It's like, you know, Darcy, all I need to do is get you for half an hour, tell, you know, I'll interview you, tell me the story, and then you're done. So, so. So here we are. I'm showing her summary on. On synthesis, and I want to make another point here. This is a great way to find out, like, holy cow. To say that Darcy specializes in small remodels that's probably 15 years, 10 years old. Clearly, her profile's out of date. So nice way to. I probably wouldn't have gone to Darcy's profile, frankly. So nice way to recognize and spot when. When we need an update to some stuff. And also just that that last paragraph is, in fact, what I'm talking about today. So I interviewed Darcy yesterday morning. If I had done this query, you know, before yesterday afternoon when I posted the results, it wouldn't have had that. So it's just. I love that we're getting up to the minute, live. This is the best knowledge that we have right now, kind of as a matter of course. Okie dokes. So here's. Here's my workflow. At a glance, it probably looks worse than it really is. I'll step us through all of this. I put kind of rough times down at the bottom for this context. If you knew, you know, a longer video, everything would take a little longer. But. But I think the key point here is this is measured in minutes, not hours and not days. This is quick, quick turnaround. So let's cut into it. Step one, and not to be pedantic, because it kind of was a big deal, is schedule the meeting. When you open up Darcy's calendar, it bursts into flames. This was her week Last week. This is kind of not an uncommon week for Darcy. I bet that the Darcy's in your firms might have a similar experience. When I was office director for San Francisco, our San Francisco office, this was my calendar. Did not like that. That's a tough place to be. So when we're, you know, talking about how come our subject matter experts aren't sharing, well, spare a moment to reflect on what their life is like, at least sometimes. Right. And so I come at this with a lot of empathy. Right. Darcy wants to share her knowledge. She is structurally incapable of sharing her knowledge if it takes a lot of her time because she's got so much on her plate. So I'm at pains to make this as easy and, dare I say, enjoyable for her as possible. So we rescheduled four times just because, of course we did right, including like last Sunday at 4pm and that even that fell through. But. But we got it done. A small, small interlude. I realized I don't actually know how to interview a person. I've never done that. But you know what? I have AI tools that can at least give me some top level, some top level advice. So that was handy. Okay, teams, we use teams. Zoom. Doesn't matter. Have the meeting. Record the meeting. Right. I think don't need to spend a lot of time on this, but here, the next, the next step is trimming. Trimming the meeting. Doing a little bit of light video editing, but it's a good place to show the timeline. And I can talk about a little bit how I structured this encounter with Darcy. We didn't just jump on a call and then immediately try to dive into the thing that I wanted to get out of her. Because you need to, you need to settle in. You need to kind of, you know, get comfortable. I need to reassure Darcy. Like, what's going to happen? What's. Where's this gonna go? Is she gonna come out looking good in all of this? You know, how's it gonna go? What does she need to worry about? What do I, what am I covering? So we spent time, 10 minutes or so just kind of at a high level. Like, okay, what is it kind of roughly, what, what do you want to talk about? What are the steps? What do we need to kind of go with this? I took notes and then I facilitated an interview effectively, right. Where all she had to do is be Darcy. And she even sort of. She even sort of. There was a moment where she's like, oh, so I don't have to worry about kind of getting to a Conclusion. I'm like, nope, all you have to do. We're just going to talk. I'm going to lead us through the, you know, what we've talked about. And you could just sort of see her shoulders drop a half an inch, right, because she's like, cool. I can just be me. This isn't going to be so bad. A thing that I have learned about doing this sort of stuff is it's really natural and easy to just have this kind of a smudge from informal chatting. Hey, good morning. What you have for breakfast? Okay, let's talk about legislative things happening in the state of California. And it doesn't give you a good place to, to trim the video to cut off that front stuff. So I always try to have like a moment of silence and then, good morning, Darcy. You sort of pretend like we're starting a thing up. So that's a lesson I had to learn. Now, hopefully, if you try this, you won't have to. Okay, back to our intranet, back to Biscuitnet. We have templates for pages like this. So gin up a template, drop the video into the template immediately. Synthesis is working in the background to generate the transcription. So that takes a few minutes. I might fill in some other, other information on the template to start getting the page ready while it, while it works. And then the next step is editing the transcript. And I indulge this process a little bit. I spend, I spend some time on this. It could probably be done quicker. My thought is twofold. One, this is the raw feedstock for the rest of this process and for, for synthesis, AI search and everything downstream. So it's worth, it's worth investing in getting the transcript dialed in. But the other, the other thing, maybe deeper than that, frankly, is as an architect, a whole lot of my awareness and professional obligation goes toward the accessibility of built spaces. And that is just drilled into me at a deep level. And it was a fairly recent awareness for me that, oh, accessibility applies to digital things as well. And Knowledge Architecture has been doing a lot of work on that because a lot of that burden falls on them as a software creator. But a moment, spare a moment to think about someone with, you know, low or no vision or not vision, hearing, who can only access this, you know, if they want to watch this video or listen to this video, sorry, watch this video without the audio. It needs to make sense because that's how they're going to access this. And so I think getting the punctuation right to me feels important because it, it makes the flow make A lot more sense. So enough on that. The transcription is great here. Heard size of the gym raids. That's pretty close to seismic upgrades. So you fix these kind of things. You know, the organization that we're talking about in this case is called hcai. Hcai. Can't blame AI for not getting that acronym right. H Space Kai is a pretty good guess, but we can go to transcription optimization and just say, hey, anytime we see that, I want you to replace it with the proper acronym. And now I find this very satisfying. That's why I. I included it. Now I never have to fix that again. Makes me happy. Okay, easy peasy. Download the transcript as a PDF and we're off to our next AI tool to generate a summary. Now, I have have say fallen in love with, but I don't like where that's going. I really like Perplexity AI. It's a. It's kind of a. It's kind of a wrapper on top of a bunch of large language models you can choose from. Gosh, six of them or so. I usually wind up using Chat GPT, but sometimes I'll play with different ones. A thing that I like in Perplexity and I think I know Chat GPT, if you subscribe to that native has this too, is the ability to create, they call it spaces. So you can have kind of like collections of. Of your queries, but you could give them standing instructions. So for this one, I have one for doing just this thing. And I say, look, I'm going to give you a PDF transcript and I want you to give me a detailed and accurate summary, about a thousand words. I probably prefer bullet lists over paragraphs, that kind of thing. So it's a nice way to make it so that every time I come back to use this, I don't have to say, don't have to give it much in the way of instructions. So here we go. Dear AI, please summarize this conversation. Blah, blah, blah, blah, blah, off it goes. Typically, typically you don't want to use the first result you get. You edit your query. Go. Well, okay, I'd like this. I need a little bit more detail here. Ooh, this is too wordy. Shorten that. I generally go back and forth for a while until I kind of get a result or, or pieces from several results, right? That feel like I'm starting to reach the sort of plateau. Maybe it's sort of 80% of where I want it to get to. And then copy paste into a word processor and do I do the final sort of hand editing of the text there. It pastes in kind of weird with the, the formatting turns into ASCII characters. So search and replace to fix that. And then, yeah, do the final editing there. Then we're back into synthesis and I'm pasting it into the page that it's already, you know, already been created. You can see it's sort of at the bottom. Most of the actual written content is cut off here, but you know what that looks like. Tune it up usually, you know, usually tinker with it a little bit, I might add. Related resources. In this case I like. Oh, we're talking about the Seismic Hospital Facilities Seismic Safety Act. Let me give a link to some information if someone wants to drill into that, so on and so forth. And there you have it within, you know, same day we went from talking with an expert to I've got knowledge resources ready to go shareable on our intranet, generally write a post to, you know, alert people that it's there and cool. And then, you know, again, just, just a touch base with synthesis AI search one more time. It's immediately available for searches on anything that sort of touches on that content. So if I search earthquakes, right. That first and second paragraph here are directly drawn from, from the post or the, the content that I just shared. Okay, that's it. Hopefully that doesn't seem hard because there's nothing, there's nothing about this. That's rocket surgery. It, there's a few steps along the chain, but they're pretty easy, frankly. And I've found it to be a really effective, effective way to kind of gin up content quickly. Zooming back out and zoom back out and we come back to systems thinking again. I talked at the top briefly about causal loop diagrams. So here's one that I think is, is relevant to what we're talking about today. And the way you read one of these is it's sort of this cause and effect loop. So the content on our Internet, as we have more content on our Internet, that increases the quality of our search results because there's stuff to find, right? The better our search results, the more people are finding what they're looking for, the more trust in our intranet as our sole source of truth, you know, grows. And the more people trust our Internet as our solar sole source of truth, well, that's hard to say. The easier it is to justify investment in the Internet to make it even better, which leads to more content on the Internet, which leads to better search results. So this is what's called a Reinforcing feedback loop. This is just a snowball rolling down a hill. This is colloquial, colloquially called a virtuous cycle. We love this. And there are some sort of influences here that start to turbocharge this. Having a slick content creation workflow that makes it just that much easier to get content on the Internet is a big boost. Having a really slick search technology that gives users a much more facile and intuitive reach into the content and brings it to them in a more meaningful way also supercharges this. So. So I find it pretty exciting. Last point on this slide. Trust. Trust is a funny word to use trust. I could have said belief. I should have said, you know, confidence. That's getting into the realm of mental models, mindsets, that kind of stuff. And I think that there is this connection. I mean really, this is talking about creating information flows. But I think that the way you move the mental models of an organization sometimes isn't. You can't sort of say, hey, I want you to have a different mental model. I want you to believe in our Internet. Sometimes that's the result of other factors influencing it. So, okay, where do we go? Landing the plane here talked about, you know, this lever of information flows as a way to move our system. But move it where? To what end? And I think the classic answer, the, the canonical, the prime directive is we shall meet strategic business goals. Am I right? And so cool. So I mean I, I would see taking this process I've described and using it for, I mean, I'm using it for a corporate university, right? So we have these kind of, you know, lunch and learns, if you will, webinar sessions. But now they are also persistent knowledge resources on our Internet with a written summary. They show up in search results. It just makes it that much more accessible to everybody. I could see using this for project knowledge reviews. I haven't done that yet. It's a, it's a, it's on my, my wish list. And we target content the way we normally do in the knowledge management world. I mean, search analytics, what are people looking for and not finding, I think will be as that comes online. That'll be a potent source of, of insight quality control data. What are our QC people tired of fixing in our drawings? Let's use that as the feedstock to curate content so that people do work correctly the first time. Quality assurance beats quality control every day. Employee surveys. I surveyed our staff earlier this year for our corporate university program. I said, what do you need to know? What do you want to Learn about, to do your job. And I got this wealth of information and in a sense it was not surprising. It was like, we want to know about design things, project management, financials, the stuff you'd think. But now I've got data on that. And so I'm laser targeting, learning content based on what people have said they want. Of course, strategic plans, so, so strategic business goals. I think that's kind of the obvious thing, but I don't think that you can start there if you're, if you're sort of entering into this. So I want to talk. I think this is my last slide. Sort of a, an opportunistic approach as a compliment to strategic approach, but also perhaps as the, as the necessary first step. So what I did with Darcy was purely opportunistic. I was ready for it because I kind of honed this process, saw the opportunity, and happily was able to take advantage of it. And that starts to move. That's this connection between, you know, an information flow like the knowledge that Darcy's sharing as a, as a lever, as a, as a place to push on the lever. But it starts to. These quick wins build credibility. They build your skills. I've learned how to interview at least a little bit, right? And it starts to fuel that virtuous cycle that I showed in the causal loop diagram. And I think that starts to move the organization not only by increasing knowledge available, but by also showing people that knowledge is a thing, that it's valuable and a thousand other, a thousand other ways. The last, my last thing here is a mental model of my own that I found was challenged and eroded by this. I had been thinking of our intranet as a place where mostly we put evergreen knowledge. We put stuff that's going to be true has a long shelf life because it's effortful to get stuff up there and to maintain it. And now I'm realizing, you know, I think the math on that has changed. I think it is within reach now to capture perishable knowledge, capture stuff that doesn't have a super long shelf life. This conversation with Darcy that I, that I, I talked about today, that's probably relevant for a year or two tops. And you know what? I am totally fine with that because it didn't take me all that much effort to get that shared and get it shared in a timely fashion. And I think that's a, that's a, a pure win. So that is what I've got. If anyone's interested, here's my QR code and there's my username on LinkedIn. And Chris, I think over to you for QA.

Yeah, Todd, that was awesome. Thank you for zooming out so so far. And then kind of going into the tactics. It was great to see that kind of range. I've got a few follow up questions for sure and I've seen some stuff in the Q and A. If you haven't added something to the Q and A panel, please do it now. And there's also an upvote feature that you can used to help get us prioritizing the questions. I want to come back to the kind of things you and I talked about in the lead up to this around you kind of perfecting the machine or at least improving the machine. And you know, I think you interviewed like a Bluebeam expert, you interviewed like a new Forma expert. So maybe start with some of those, how you got to those lower stakes kind of things before building up to someone like Darcy who's hyper busy as you mentioned.

Yeah, that's right, that's right. I wouldn't have tested this for the first time on Darcy because the cost of failure would have been bad. Bluebeam. Yeah, actually it was a technical. It was a very software picks and clicks thing. We have. Not only do we have a Darcy, we have an Emily. And our Emily is a Bluebeam wizard. She's amazing at it. And guess what? We don't have on our intranet, we don't have much content on Bluebeam. And that's been a request. And so I just, I kind of, yeah, I had a notion of like I want to try something and, and Emily would be a safe person to try this with. And so it was in my judgment a reasonable, you know, like worst case scenario. I will write this by hand based on the recording. Right. Best case. Holy cow, I've figured something out. And so, so yeah, so doing that and then doing the same thing with our corporate university presentations, we do these monthly now and kind of honing it on that process where that case, no one was really expecting me to do anything different than just post the video. So it was pretty low stakes way of, of kind of working out the kinks.

What was the biggest takeaway that you learned? Like what was the biggest kink you had to work out?

I mean there's little things about sort of the tool chain. Like Perplexity loves to take a PDF. It will not accept any kind of word document or like text file. But that's, that's kind of fiddly. I want to have a better answer than the One that I do, which is this. Having a good, mindful, like, clean start to edit the video. Video editing sucks. It's hard. And making that easy by being mindful about working with a presenter if it's a PowerPoint and saying, okay, I'm gonna do some housekeeping and then I'm gonna give it to you. And I want you to take a breath before you start talking, because that's where I want to cut the video.

Yeah, yeah, Makes sense. Makes sense.

Just some facilitation skills, frankly. Yeah, yeah.

We had a question kind of going completely different direction, but it's kind of getting into your virtuous cycle a little bit. Bill Spalding wrote and asked about thinking about the new search if you've had to remove data that's outdated and correct or update data like, tell us about kind of. That kind of quality workflow.

Yeah. And Chris, I might kind of give this back to you. So in the beta right now, the way it works is not. It doesn't wait. It doesn't have any kind of awareness of how outdated or old something is. And so it treats everything on our site kind of equally from. To my. To my. To my eye, Chris, I understand that that will change. So the way it works for us in the beta today, there's like, oh, geez, I need that ancient stuff to go away. I don't feel like that's going to be a problem going forward.

Yeah, I mean, I think it's a good question. And you're right. We have not tuned the weighting. That's obviously going to be in the pretty high on our priority list in the next version. But there's still going to be a challenge, like as much as we can say, prioritize a post from 2024 about seismic upgrades versus one from 2014. There is a question. If the only thing on your intranet is from 2014, it could be that it's true. We don't know. Just because it's old doesn't mean it's wrong. And so there is. The algorithm will help when there's competing content, but when it's the. I'm not actually going to respond from you. I'm kind of saying this is kind of where this is going, is we'll try and cover up as much as we can when it's obvious there's more relevant or more updated information. But if that's the only thing you have, it's like, are we suppressing something? Yeah. Anyway, you get it.

You know, there's a degree to which I See this as a feature, not a bug. If this is. If this is a way that we start to notice, okay, we have some stuff that is. That is outdated in a way that's unhelpful, but now we can detect it and spot it. Well, now we can manage it, right?

Totally. I think one thing that I've seen you doing and other people in the beta, as well as our team who's using it, is they're using the rating feature to basically say, this content's old. This is why this search didn't work. This content's old. I'm going to go fix it. And so that's a very useful workflow that you guys I'll learn about soon.

Yep.

Jim Martin asks if you found a way to get AI to do the interviewing part instead of you.

Is that. Is it. Is that just Jim being funny? I don't know. No, I haven't. I haven't. And it wouldn't be high on my priority list because I think it's fun. I think that's. I think that human interaction is nice. I think too, that. That there's a. It is a lot of what I'm trying to do with this is make it not weird for the expert. They've got so many reasons not to do this. I want to remove as many of those reasons as I possibly can. So having it be uncomfortable, the discomfort of writing Right. Is such an impediment. Well, cool. Take that off. Off the place. But, hey, Darcy, I want you to talk to this AI in the void. I just don't see that working, you know, when I. And maybe, maybe this is a better answer to your question about kind of, what have I learned? What insights have I gotten? When I did that, that trial run with. With Emily about Bluebeam, one thing that I thought to do was get her an audience. And so I invited four or five folks that I figured could use that knowledge like. Like that live. Like their time would be well spent, you know, in the live event. Because now she's talking to people who are receptive to the information, and that feels natural and she's delivering value in the moment. She's not speaking to a camera, you know, by herself in a room, feeling like a weirdo, wondering where this goes. And so you have them ask her.

Questions too, or did you control. Were you the only person asking questions?

I let it be organic and just. Yeah, I figured the questions kind of add to the. You know, they're going to clarify things that are unclear or whatever. So I didn't I didn't facilitate that much as I do this more, I see the role of the facilitation, you know, as a. As a pretty important piece to it, to getting just higher quality video.

Did you see one of the kind of things that if you combine your kind of lever and also your kind of virtuous cycle, if you take a different perspective on it, that from the lens of the SME, there could be a mental model that Darcy or Emily or people are carrying around that's kind of like shame around sharing knowledge. It's like, I know I should be doing this, but I haven't been. But the burden feels so high. It's just like the shame, vicious cycle. Whereas now, like, I wonder if you've seen shoots of like Emily wanting to do more of this with you or others because they know that it's actually easier and impactful and they can get out of that loop.

Yeah, yeah. And that's a little bit what I was getting at about, right? Like shifting the mental models, you know, not by pointing at them and telling people to change what you believe, but by like showing what's possible. An honest answer is to be determined. Right. Because I've only been doing this kind of couple of months and. But already I'm seeing if I, if I sort of abstract a little bit. The more we. That that virtuous cycle of. The more we invest in our intranet, the more people are like, oh, hey, I have something that should go on the Internet. I'm seeing that lighting up a lot. And I just. And in a way, I see this specific thing I'm talking about as an accelerant.

Yeah, yeah, that makes sense.

But you know that the next time I am talking to some subject matter expert in our firm and they kind of go, I know I should share this. I'll say, have you seen the video? The thing I did with Darcy? Right. Talk to Darcy. This was really easy and fun, you know.

Oh, that's where you go. Yeah, yeah, yeah. So there's like a benefit to starting with lower stake stuff. But if you get somebody senior like Darcy, then you can also point and say, like, you know, so you've got like a reference, you know, at a very high level that will kind of.

Right. There's. There's a quick tangent. There's a book called How Buildings Learn that I read forever ago and has this example of a person builds a house and then like they die or whatever. Their kids are like, you know, it'd be nice. Be nice to have a porch. And then Their kids are like, you know, it'd be nice. We need another room. So they wall in the porch. And then their kids are like, you know, it'd be nice. We'd like to have a porch. And so like, everyone sees one step down the road and I think that's kind of true. So I had an, you know, a couple times now where something came up where it's like, holy buckets. This is a moment where I can do a full dive, like transfer critical knowledge. Dorothy Leonard. And that was. That was walling in the porch. Before I built the porch, people were just like, I don't even know what the hell you're talking about. And they would sort of back away slowly. Yeah.

Do the simple thing first kind of thing first. Todd, thank you so much for your generosity. Not for your time now, but preparing and kind of bringing this. This is a. We're in a whole new era now with knowledge management and I think people like Euan that are pioneering this and sharing with the rest of us will help us all get better. To that end, small plug if you didn't see it. K Connect 2025 we announced registrations open as of this morning. Our theme is unlocking the promise of Knowledge Management with AI. So if you like this topic, we're going to be spending a couple days talking about it next August. And Todd, thank you again. Thanks everyone for attending in your chats and your questions and I will see you next time.