The AI Readiness Report: The Construction Leader's Journey from Curiosity to Confidence

My name is Randall Mauricio. I’m the fractional Chief Operating Officer at Tag CXO. My role here is to strengthen the operational backbone that allows TAG to deliver its expertise through its enterprise caliber of fractional executives and deliver that service consistently and at scale. I’m going to click forward to our agenda today.

(02:12) A couple of things we have on the docket. We have obviously where we are in AI, where we’re going in the future. But before we get into that, let’s go into some introductions.

(02:24) I’d like to introduce our two speakers today, two gentlemen who have oodles of history together. I’ll start first with Paul Theisen. Paul Theisen is founder and CEO of TAG CXO. He’s an executive stewardship firm helping middle market companies turn C-suite leadership into structural advantages. He’s a former three-time CIO, and he brings 28 years of corporate leadership experience across commercial, construction, and global manufacturing. Paul founded TAG CXO to give growing organizations access to battle-tested fractional and interim executives to bring clarity, judgment, operating discipline to complex challenges. And he’s the author of a forthcoming book titled Outmaneuvering Goliath. And Paul, this is something we’re excited about to formally announce. Just in a couple months now, right? What are we targeting?

(03:23) Yeah, yeah, a few months. The book doesn’t come out until late Q1 next year, but it’s been my passion project. So I’m excited to tell the world about it.

(03:26) And Eric Sanderson is the founder and president of Red Rocks Advisors. He advises engineering, construction, and infrastructure organizations on strategy, management, project partnering, and of course, execution. He has more than 25 years of industry experience and Eric combines a broad view of business strategy with practical knowledge of field and operating realities. And his work, of course, is concentrated in utility and infrastructure environments, and that includes water and wastewater projects. He’s also an IPI certified master level partnering facilitator.

(04:00) And Eric has led more than 325 professional facilitation. So when I say these two gentlemen are highly accomplished, highly experienced, we really, really mean that. And I’d love to share just some background context of their two relationships. These two gentlemen have been have had enjoyed a professional relationship for nearly 20 years. So that’s a long time they’ve worked together as co-collaborators on many projects in the construction industry, including field automation, learning and development strategies, as well as business operations and assessments

(04:33) And their latest partnership is the reason we’re all here today for this live event, and they’ve called it, they’ve dubbed it the AI Readiness Report. And I should emphasize this is an ongoing industry research project

(04:46) And so, yes, there’s value today, but there will yet be value in the weeks, months, and years ahead. And all this is to lead and facilitate the discussion on AI adoption and construction by talking directly to the stakeholders inside the construction industry. And their goal simply was this to discover where we are, who is doing what, where value is starting to emerge, and of course, what risks and headwinds are present today and in the future so that industry peers can navigate this new landscape.

(05:18) So, with that setting of the table, I’ll toss over to our two speakers today. Thank you both so much for taking the time to put this presentation together. I will continue manning and leading the presentation slides, and Paul, I’ll toss it over to you. I will follow your lead.

(05:33) All right, very good. Thank you so much, Randall, for that. And welcome, everybody. It’s a pleasure and a privilege to have your audience today. So thanks for being here. We know your time is valuable. We hope you get some benefit out of today. So why the AI readiness report? Eric and I have had a lot of conversation over the last few months, and this idea sort of came to fruition when we recognized that we need to hear from the industry, and the industry, the construction industry needs to hear from their peers. And so we wanted to create a safe place where people could come and learn from one another and what’s going on across the industry with their peers and to create their own situational awareness and start to formulate their own plans. So this is really about situational awareness, about having a safe place, and it’s about hearing from your peers. So, Eric and I are really just reporting back what we’ve learned from you all, so thank you for participating in the survey.

(06:29) This slide right here, this is not part of the survey. This is part of the conversation that Eric and I had when we were going through the survey results. Look, this is a four-year-old, not even a four-year-old phenomena that we’re all being exposed to, where this is widely… been widely available

(06:48) And in fact, here’s just some interesting facts about this. ChatGPT, which sort of started this phenomenon, November 30th, 2022.

In five days, I’m sorry, in five days, 1 million users. What immediately hit the news cycle in 60 days, in two months, it hit 100 million users, the fastest new technology adoption in the history of the world. So, and here we are, a group of leaders in the construction industry being asked to figure this out and govern it. And what do we do with this, and how do we apply it to our businesses

(07:30) Let’s give ourselves a little bit of grace. Let’s take a big deep breath and offer ourselves a little grace here. This is a lot to consume. We’re drinking out of the biggest fire hose that technology history has ever turned on us, and we’re all here to try and figure it out together. So

(07:47) You know, there’s been some AI development historically, as you can see from companies like Autodesk and many, many others, but nothing like the generative AI revolution. And of course, Gen AI is not the only thing that AI is. As we’ll find out, there’s assistive AI and there’s Agentic AI and what are those things? But look, this is a very new phenomenon for those of us leading in the construction industry. So have some grace. We can move to the next one.

(08:19) And this is really the takeaway here. Industry is being asked to establish governance training and some return on investment for technology that’s only been broadly accessible for barely a few years. So, that’s really the takeaway here. We can move forward. And Eric, I think we could turn it over to talk about the actual survey respondents.

(08:39) Yeah, absolutely. Well, good morning and thank you for letting me be a part of this. And I recognize a lot of names that we have in the attendees, so I appreciate you all taking the time out of your day to sit in on this. This has been, as Paul had alluded to and Randall shared with the introduction this initiative has come up relatively recently, just in our work, and almost every single meeting I’m in lately with clients and the executive roundtables that I manage, the conversation shifts really quickly to AI, what’s being used and how it’s getting used, etc.

(09:11) And so, in some conversations recently, Paul and I were like, we need to be addressing this. We need to hopefully be able to get our arms around what’s happening, and maybe if there’s any guidance or advice to be able to be provided to identify that.

(09:24) So, as you can see, we sent out the survey. I appreciate, again, everybody that responded. We had what I think is a tremendous response rate based on the timing of this. As you all know, anybody that’s on with us right now, and you responded to this survey, this is only 2 weeks old. This is very fresh data. It’s very real

(09:41) Great. It’s very real time. So I’m kind of excited about that. Here’s the industry cross-section of who responded. 47% on the GCs, and then we see a mix of other types of companies engaged in the industry. What this really… what this tells us, and a couple of things, is that there are probably more applications for AI than we even can contemplate right now. The initial pass through, I think everybody maybe on the call was thinking, hey, this is something we can use to improve how I write an email so it’s not quite so snarky, right? But suddenly we’re seeing that it has started to touch all aspects of our industry and all types of companies engaged in our industry are looking at it, saying, what do I do? How do I do it? And some are they’re a little bit more advanced than others, but it’s a good cross-section, both of respondents as well as some discussion around application.

(10:36) Let’s go ahead and move forward to the next one. The, continue with the cross-section of respondents. This, I think, is very enlightening as well. Again, I want to thank everybody that responded, because what this gave us, almost more than anything, is clarity that this is a full industry perspective that we need to be thinking about here. This is not a challenge or an opportunity that is exclusive to one sector of the industry or a different sector. You know, from the smallest companies to the largest, the respondents, I mean, we really hit I think a good cross-section of the entire industry. And it’s fitting with the type and size of work and size of company that we’re working with. If we see over here between, say, over $500 million in revenue companies at 28%, I can tell you that large companies like that and even going to the number of employees are reflected, I would say, it probably disproportionately represented in here, because there’s a lot fewer of those very large companies, and there’s a bunch of smaller companies in the industry. Again, what this is showing us is that we’ve got a really good cross-section of respondents, and so I feel like we’ve got our finger on the pulse right now of what’s been, you know, what, hopefully what’s been happening, and some of the thoughts and concerns about AI and how we’re beginning to use it. But yeah, great responses, and again, thank you to all that did that.

(11:59) The markets represented gives us a little bit another piece of information I think is very interesting here. Obviously, we see a spectrum, but it is not, again, exclusive to one area. Commercial, if one thing, you know, a number of respondents hit commercial, but this was a question that allowed multiple answers to the same question. And so, it’s pretty predictable that sort of the commercial is your catch-all, and industrial being right behind that. But what we’re also seeing is that there’s a lot of attention, a lot of focus from the more specialized contractors and industry providers out there within the utilities, water as well as energy. It’s definitely stepping out. And then manufacturing, you know, that differently how do we use it? What are we looking for AI to do? How can it help us? You know, again, that’s a little bit different application we might see in regular contracting, but again, it’s part of the overall mix that feeds the industry.

(12:51) There is one other piece I would want to draw out, and I’ll speak to this a little bit later as well, but we’re also looking at, there’s a reasonable percentage that responded that is doing federal work and when we start to talk about different types of work that the respondents are doing, what this is going to speak to is potentially different requirements of how you use technology and how we maintain compliance with requirements per contracts, et cetera. So each of these different industries or sectors of the industry has different opportunities, different needs, but then also different risk profiles on how we’re applying AI in the industry. So this is a very… I thought this was very insightful, and we’re going to dig deeper on this as we move forward.

(13:38) Yeah, a lot of good information in here. Yeah. No, no, please. And as we dive into the findings, sorry to interrupt, Eric, just a quick reminder to the attendees here, feel free to use your chat function in Zoom and of course, as we dig deeper into the findings, if you have a question that crops up, please use the Q&A function. We’ll be storing those Q&A, pardon me answering the questions during our formal Q&A session on the back half of the live event. Over to you over to you, Paul.

(14:02) All right, thanks again, Randall. Okay, AI use is already widespread. This is just an interesting and early finding that we thought we would share with you. This slide basically says everybody is using AI, and nobody is not using AI. So if you look at that to the far left, 90% of the people using it weekly, those same people, 69% using it daily or multiple times a day. It is part of everyday life. And oops, yeah, go back to that, go back, just… there we go, okay. Say their organization is not yet using… so, yeah, almost nobody’s not using it.

(14:42) Well, if I could jump in real quick and offer a quick comment. What this also me is so many people are using it and it’s being used so broadly that it almost, we’re going to move from just using it to say it’s interesting to somewhat dependents and reliance on it. And as most organizations are starting to embrace it, you know, are they going to lean into it a lot more, probably than anticipated.

(15:05) Yeah, yeah, right on. And actually, Paul, we do have our hand up from Linda Fulmer. And so if I could ask our back end moderators to allow Linda the opportunity to share a comment, or perhaps even a question that might be an opportune time to do that right now. Sure.

(15:24) We’ll just give our team a moment, a couple seconds to coordinate that. And Linda, if and when you’re able, please do. Oh, one to decline. Okay. You know what, Linda? It sounds like in case there are some tech issues, encourage you to use the Q&A function. We’ll make sure we get your comment and question to our speakers very shortly. So back over to you, Paul.

(15:44) All right, very good. Yeah, so clearly adoption or everyday use is way outpaced the readiness of organizations. And we’re going to come back to that issue. We’re going to hit on that issue again, what readiness really looks like. So yeah, let’s look at the next slide

(16:01) Today’s use categories versus tomorrow’s anticipated value. So this is what are people doing today on the left? It’s all personal productivity related. Meeting summaries, emails, proposals, you know, personal research. This is all things that I need to do personally or in my role that benefit me personally. But that’s not where I see the value. This is what you guys have told us. We see the future value being in business process automation-related categories, like estimating and project controls and contract review and and the like. So, there’s a huge gap between what we’re doing today and where we think the value actually is. You know.

(16:47) So that’s a… that’s a key takeaway. Eric, anything to say on that, Eric? Yeah, what jumps out real quick here is our early, what we’re seeing in early adoption is really what I would characterize as low hanging fruit.

(16:57) You know, meeting summaries is, you know, we’re recording stuff, it’s producing some notes, emails, cleaning up, you know, verbiage and things of that nature, and proposals, you know, helping position value propositions, etc. Research, pretty straightforward. So, that makes sense. It’s low-hanging fruit, and it’s… I’m gonna argue it’s relatively low risk. Right. Right

(17:18) As we get into the next piece where organizations see value, I would characterize that as the heavy lifting that people are looking for. Truly some assistance with improving processes streamlining, you know, the level of effort, seeing reduction in time and how we’re using our time for regular business processes. So, what I’m seeing is that organizations want to see AI being used not just for that personal productivity piece, and not streamlining, but truly improving how we execute the business. So the expectations are high, that’s what I’m seeing.

(17:51) Yep, this is it. This just tells the classic story. Nothing ventured, nothing gained. You know, I think there’s some small personal gains for everybody, but there’s huge potential business gains, but it’s going to be a much heavier lift. Yeah. Yep.

(18:07) Absolutely. I thought it was a very interesting question and we’ve got some amazing responses here. I think it was very insightful also. Barriers. What we observed here is that a lot of these barriers are leadership-controlled. They are, they’re areas that are controlled by strategy, the ability to execute and do it in a disciplined and risk-controlled environment.

(18:35) The interesting thing that really, obviously, is going to jump out here that everybody can see here, is that cost is not any sort of a barricade here. It just simply is not getting in the way to be able to use this. I had a… as I was preparing for today, one of the things that jumped out to me is I thought, what if AI use and be able to access it on a regular basis was super expensive?

(19:00) Would that change what we’re looking at and how we use it? And I would speculate, of course, yes. But now we have a situation where we have these incredibly powerful tools that are very low cost and very cheap to get into and to be able to use. And I couldn’t help but think about this as almost like a like a rental car. This is going to sound funny, but I can’t help but use a quick analogy here. It’s almost like renting a Ferrari for 20 bucks a day, right? But doesn’t come with a map. There’s no destination and there’s no traffic laws. Yeah.

(19:37) And so what we have right now is cheap access to incredible power. And what do we do with it? Where is it going? And I think inside of that, that starts to create some, hopefully that might be a little bit of an analogy that begins to make sense that we have this, again, tremendous power available to us, and it doesn’t cost much, but then we have all of these secondary sort of barriers and implications, and it’s not about how we’re using the tool, it’s about how we’re managing that use and governing it. So, yeah, I thought this was very interesting. There’s a lot more to unpack with this. Yeah, yeah.

(20:10) So here we go. This is kind of continues that same thought is that the access to AI is not about capability. It’s not about skill set. Interestingly, when we see who is paying for AI licenses throughout the variety of different different models that are out there. 84% are being paid by the companies, 74% of AI is self-taught

(20:34) I think that’s very interesting. So the companies are paying for something that employees are teaching themselves. Understandable, right? Because it’s very organic, it’s very grassroots to a great degree. But this is where I think we start to begin to realize potential risk, is that 64% of those characterize your knowledge base at a beginner level or somewhat limited. Right? So we are essentially, you know, handing the keys off, or giving keys access, you know, fill up the tank as much as you want type of thing, to somewhat inexperienced drivers, right? Now, we don’t have a lot of mileage here, so I don’t know that we could say any driver in the industry is experienced. Yeah.

(21:25) But it would be the equivalent of handing that $20 a day Ferrari off to somebody without a driver’s license. And that’s kind of what we’re looking at here, is that there’s a lot of opportunities, a lot of speed, a lot of power, but organizationally, we’re not seeing it well supported across the industry. And that is further demonstrated in that last little graph there on the right about having an AI champion or committee at only 34% that we know, again, let’s go back 97% of the industry is adopting or using AI in some capacity. Only 34% have any level of control over that. That’s a big deal. Paul, yeah. And their employers are paying for it

(21:56) And their employees are paying for it. Yeah. So Yeah. Yeah. You put three things together plus 90% adoption, right? 84% of the companies are paying for it, but only a third of the companies are actually managing it.

(22:07) Yeah, yeah. And so that leads us to sort of a landscape that says there’s some pretty substantial risks here that I think bear consideration and bear some dialogue. So we’ll go ahead and roll to the next one there for some input.

(22:22) So this is kind of interesting as well. The concept of governance, it’s sort of how do we ensure that what we’re using is being used appropriately is being used right, and is not putting the company at risk. This is, I think, where where we need to have some, you know, good and important and serious conversations. Everybody by now has heard of the concept of hallucinations, and as, you know, produced, generated some type of output from whatever AI tool they’re using, and has had the opportunity to be suspect of its accuracy.

(22:54) Did it make up something entirely? Did it fabricate information? Or is it simply needing some accuracy checks? And I think right now in our industry, what we’re seeing is that there is a big concern about that, and this is probably one where we see the most evidence of the risk in the industry, there’s a lot of anecdotal stories. I hear them with my clients and I hear them with outside in the industry as well as other industries that leaning in on the reliance of AI too much can create problems that can lead us down a path that doesn’t, you know, doesn’t work well. Now, some of that is mitigated through, say, for example, pay models. You know, if we’re behind the paywall, in theory, we get a little bit better accuracy and quality.

(23:42) Some of that also would be where does AI sit relative to our firewalls? Are we using some public access AI to where we’re going out? Maybe we pay, and in theory, it should be better. But in some organizations, we’re bringing it in-house and downloading and using something that is a little bit more sequestered. There doesn’t necessarily guarantee accuracy, or it doesn’t guarantee that the information that’s created, it’s generated, is 100% reliable. So it still requires a fair bit of human interaction with and human contact to verify that the information that we’re looking at is accurate and correct. And as we get into, you know down the road, we start talking about use cases and where in the business that you can begin to use it, that concern about hallucinations and accuracy, I think, goes up tremendously. So

(24:30) You know, when we’re dealing in the early adoption around, say, wordsmithing, proposals, cleaning up emails, those sorts of things, those are low risk, right? As we move forward and we start to apply AI to hard science, mathematics, you know, anything you know, estimating, job costs, things of that nature, the accuracy becomes a much more critical issue.

(24:54) Three other… I’m sorry, if you don’t mind, pop back on that one real quick. Three other pretty significant concerns that we’re seeing, and in fact, obviously, these are bigger concerns than the hallucinations, is that data leakage, intellectual property loss, and then data security.

(25:10) These also become fairly important depending on the type of work that we’re doing, and I think that this is something that we’re ferreting out right now in the industry, is what is the risk when we are uploading data to a model, even if it’s a pain model, but that is outside of our firewall. You know, everybody on the call here or on the webinar, you know, these are well-led organizations, and you all pay a lot of attention to your cybersecurity protocols and suddenly we have this tool that kind of weaves its way back and forth between proprietary information and then sort of public domain stuff that gets lost, frankly, once stuff gets uploaded, largely if it’s outside of your system, you really don’t have control of it anymore, even though we may be using a pay model. So, there’s a lot to be discovered, a lot of research that we still need to do here. I think it’s moving us down that path.

(26:04) And then on the right side, we’ll see, this is interesting again via the survey, the governance gap that we’re seeing. I mentioned that previous slide we had 34% have an AI champion or committee, only 33% have any sort of an AI policy. So now you pair that up with the amount of usage, 97% usage in the industry that at least are aware of, but probably at a personal level, it’s potentially closer to 100 without a policy, without any guidance at all that says, don’t upload this, don’t use this. Don’t download this. Those sorts of things we would consider to be fairly basic, but not having that is represents a risk in the industry

(26:46) And then the one down below, 23% of the respondents didn’t even know if their organization has upload rules related to AI. So again, now let’s go back and kind of think about what’s that cross-section of respondents to our survey. And we’re seeing it’s across the industry. We’re talking about, you know, thousands of employees, hundreds of millions, billions in revenue of organizations that are involved. And we have a chunk of our industry that doesn’t even know what their company policy is or if they have one at all

(27:18) So that’s where I think we’re starting to see some red flags, some items that we need to be paying attention to. Yep, right on. This is me. All right.

(27:29) This is you, Paul, yeah. Yeah, so I just… I love listening to you wax on, Eric. All right. Yeah, this is such an interesting finding, I believe. Investments will rise. Everybody, well, not everybody. The majority of people, right? Three quarters of the people that responded said, yes, investments are going to go up. We anticipate spending more money in this category.

(27:53) This is important to us. Of those respondents, they don’t have a strategy and they don’t have a budget, there’s nothing formal. So it’s an interesting time when we’re expecting spend to increase and we almost welcome it, but it’s outside, there’s no real identified budget yet and there’s no real identified strategy. And as Eric was just telling us

(28:18) There’s no real strength of management infrastructure or controls or governance yet. So, I would think that a large part of the spending increase needs to be governance-related, not just tool-related. So I think that’s a key takeaway.

(28:35) You know, I think about times in our industry when credit cards were made available to a broader staff, right? There was a time when everything was sort of kept tight, but then recognizing that we needed to spend money in the field periodically, you know, then we issue credit cards. And that got to a spot where I was like, wait, wait, wait, wait, we can’t do that. We’ve got to pull it back because the pendulum seems to swing back and forth. And what we saw on a previous slide, 84% of companies pay for those user licenses, but 64% have no no budget. So right now, we’ve got a basically a blank checkbook with unregulated access. So it’s not blank entirely. You know, I recognize everybody’s not going to let somebody go spend, willy-nilly, but it’s just interesting that this tool, it’s very Very exciting, but there’s some controls issues, right?

(29:25) Yeah. Yeah, a lot of excitement. All right. In terms of just a general watching our clock. So FYI for Paul and Eric, I think we’re making good time in about eight or more slides, we’ll transition to a Q&A, which sets us about on par to start our Q&A session in about 10 or so minutes.

(29:42) Perfect. Thanks, Randall. Okay, very good. So, at the risk of beating a dead horse, I think we’re starting to come in for a landing here. You guys can read this, you know, you know, the challenge is not adoption. Everybody’s using it. The challenge is how ready are organizations to get value out of this.

(29:58) And in the next phase of this is leadership, management, governance, coordination, controls, policies, and some validation and some auditing that things are running properly, and some measuring that there’s actually value being generated. And so there’s definitely a gap. So we can move to the next one. This next graph, just a little bit busy, and to be honest with you guys, I was doing my notes, and I developed this, and as trying to understand what the dynamic is and what’s going on. If I had to do… if I could do this over, if I had time to edit this again before the webinar, it would look like this. Are we ready, willing, and able? I would add ability to this, the classic phrase, ready, willing, and able.

(30:46) And we are in the upper left right now with the situation that we’re in. It’s very low readiness, very high adoption, daily use. We’ve beat that horse today, and that’s construction in the age of AI. We are willing. But we’re organizationally not really ready. And as far as governance goes, we don’t have that ability yet. We don’t know what that looks like yet. If you follow that line, which is flip the, you know, which is basically flip the script on construction, go down to the lower right, this is the traditional, I was a CIO to the industrials for many, many, many years of my corporate career and help a lot of industrials now, you know with fractional executive services. I would characterize construction over the years as low tech, low touch, late adopters.

(31:37) So the tech is ready and the governance infrastructures are there to manage the technology, but the people largely have been unwilling over the years when you compare it to other industries. And so for this is completely flipped the script on that. So obviously we need to take AI from the upper left to the upper right, which is where all technology projects and technology implementations and technology adoption really needs to be.

(32:06) We’re ready, we’re willing, and we’re able. So we’ve got the willing, and we’re even going to put money, there’s interest, people are interested, there’s money, there’s budgets. The technology is the technology ready? It’s good, and it’s improving, and it will continue to improve.

(32:22) The biggest gap right now is corporate or organizational ability, so it’s governance, okay? So let’s flip to the next

(32:31) And, you know, in doing all of this discussion and research, Eric and I sort of I had to frame up what is the maturity path here, and this seems to be the logical path that people are taking, and maybe this is insufferably simple, but it really tells a pretty important story. I know this is busy, I apologize for that, but if you start at the left, it’s me. It’s what I do personally in my job. It’s what I do personally with my team, and then it flips to our business processes and then to our entire company. So we’ve sort of turned this into a maturity model that we think makes sense. And again, if you look at the very bottom here narrative, 92% are using it, 60% use it daily. I think that’s actually 69%. And if you go all the way to the right, it goes down to 11% have enterprise-wide AI initiatives. So, it’s… everybody is on the left-hand side of this continuum right now.

(33:31) And we’re slowly need to move to the right of this, because why? Because you say from the survey, our processes and our company overall is where we’re going to find the value.

(33:46) This is where we’re going to find value. So this is that’s the work that’s ahead of us. So

(33:52) Paul, if I can comment real quick on this, traditional implementation of technology actually flows the opposite path, right? It starts as a strategic initiative to say, hey, we’re gonna implement this, whatever that might be, and then it begins to move down into, you know, subsequent groups within the organization, and finally

(34:10) You’re right. It moves down the org chart instead of up the org chart.

(34:13) 100%, 100%. This is so dramatically unique in our industry to have something that is organic. It is coming up through. It’s like grass growing up through the floor, right? It’s there

(34:23) Yeah, but it really has flipped the script. Yep

(34:26) Yep, 100%, 100%. Now, the other thing I want to highlight real quick is that the behaviors, the current adoption model, what we’re seeing is the heavy on the individual and then application at the job level. And so in that space, we develop habits. We develop, you know, behaviors that say this is how I’m using it. But what happens, the risk I think we’re going to see is those habits, those behaviors carrying through to strategic implementation. And some of those habits, some of those behaviors may not be the best

(34:57) And speaking very specifically around security, confidentiality, and some of the risks, traditional cyber risk components that a company would face. So yeah, it’s an interesting dynamic right now seeing this script flipped around here. Yeah. Thank you.

(35:22) We do have a question from Albert Liu, and I’d love to pose this question to Paul. I’ll recite it so that, Eric, you have the ability to hear it and weigh in as well. But here’s a question from Albert.

(35:23) What are the ways to measure value for AI adoption? Traditional ROI measurements do not often work because AI spend does not always directly translate to more new business and more revenues, etc. Paul, we’ll start with you first and we’ll go to Eric afterwards in case he has any commentary.

(35:40) Well, first of all, hello, Albert. It’s nice to hear from you and it’s been a while. Look forward to catching up. And second, could you start with a harder question, Albert? I mean, seriously. You know, I really believe how do we find value? I think organizations need to come together with business leaders, business opera, operations people, where do they find friction in their businesses? Because that’s where the margin leak is. Where are they finding friction? Is it in the estimating process, in the submittal process? You know, where are the core processes where you’re finding friction. And what would it be worth to you to start to shore some of those up and start to automate those? I don’t think it’s all that different from traditional automation, but I think the leverage is different. And then I think, go very narrow and pick one or two pilots and start to rigorously test those. And prove it out, and work with that until you understand what questions you should be asking and what surprises you’ve had, and work through those situations and some of the disappointments, and then figure out how you’re going to take it to the next step. So prove out the value model before you overcommit, right? So, identify the universe of things that you think you want to do, narrow it down to two or three things, pick one or two of those, and work on them until you find some value. And then, once you have an organizational rhythm, you can start to gain some momentum, and put some more budget dollars on it. But it needs to be coordinated. I guarantee you, the biggest margin leak is not having somebody accountable. You can’t just have a Wild West. Pick some things, agree organizationally to work on them, and then control the testing and the variables and the testing and prove out the model. At the end of the day, the business needs to answer the question, does this bring value to me and my teams. And do we think this creates a benefit in the organization?

(37:44) Paul, if I could chime in real quick as well, this is a really good question. And it what everybody wants to know. How is it going to create value for me? How do we measure it? I can tell you there’s a course you could say, does it translate to new business or more revenue? It depends how we’re using it, right? If we’re using it in the business development process, then we measure the top line.

(38:02) Where I think that we can see a lot of opportunity is really in the direct costs, right? You know, that translates, of course, to the bottom line, which is the most important piece. How do we improve a business process? Make something up. The cycle time on getting pay applications submitted, the accuracy checks on those, support documentation, et cetera. How much time does it take to put together a pay app and get it submitted? Can we streamline that cost, reduce the labor hour time that it takes to produce a cost, or I’m sorry, a pay app. That’s measurable. So what that would say is that how do we measure that? You pick a process, as Paul had talked about, but you baseline it. What is it costing us now? How do we do this now? How much effort is involved in this process? How do we apply AI? Are we using it specifically to improve a process? You administer that, you measure the difference, and there you have some results. The other aspect or another possibility is not just sort of the cost structure, but the accuracy, right? It doesn’t speak to the cost directly, but it does in terms of the amount of rework. When we produce more accurate information, we’re spending less time fixing it. And then the last piece for measuring it is, does it reduce risk, right? And that’s a little bit maybe harder, because risk unmaterialized. If it’s not real, we haven’t necessarily created a cost impact, but risk, when it materializes, is dramatic. So I think there’s a couple different ways that we measure that value, and the ROI does not always come from the top line, and frankly, the top ROI, top line value is not what matters. Bottom line is what matters.

(39:38) Yeah. Boy, Eric, that robot. Thank you for submitting. Yeah, yeah, that’s really, really well said. Eric, that reminds me of some of the work that we’ve done together where we actually studied it was north of 25% of the work that came from the field to the office required rework and eliminating that is a huge source of improvement. So there’s soft benefits, there’s hard benefits, there’s soft costs, hard costs

(40:01) Sure. Yeah, absolutely. Okay. Thank you for submitting that question, Albert Lou and Paul Eric all direct our attention back to the slides. We have about five more slides to get through, and then we’ll come up for Eric to answer any more questions.

(40:16) All right. You’re doing great. I will try to keep it a little more concise. So what we’re observing, and Paul had articulated this earlier in terms of this maturity progression, we see that it begins to emerge and it becomes real plain as day.

(40:29) What we see now is that obviously AI is getting used and the innovation, the ways to use it are so broad that it’s very exciting, right? It’s very compelling. In fact, our industry is somewhat unique because there are so many different attributes of our industries, production, there’s administrative processes. There’s, you know, estimating the business development, all… there’s, you know, HR functions. There are so many different dimensions that AI could be, you know, useful in that it’s almost, there’s too many choices. It’s like going to a restaurant, and the menu’s so big, or is it 10 pages? Like, I just really wanted one thing, right? So that’s a little bit tough. That’s a little bit tough right now, is that we’re seeing a lot of use, but there’s a gap in that use. That gap is that lack of strategy, lack of prioritizing the best place to position use of AI. And we’ll talk… we can talk about funding, but really, this is the central observation that we’ve come up to so far is governance, that this tool, it’s powerful, everybody’s using it. There’s a lot of excitement, but is it being governed? Is it under control? Do we run the risk of it getting a little bit away from us? And that’s where I think what’s happening now, there’s a lack of leadership. Where do we need to get to? That leadership has to create the strategy. It has to create the path forward and put some bumpers, some guardrails on this.

(41:53) I think, you know, in our industry, we’re gonna see people will want to embrace and move forward with that, and sometimes a little frustrating when something exciting new and suddenly we’ve got to put some controls on it. But I think the risk profile that we’re seeing demands that this needs to have some guardrails. But when we put those guardrails, we’re actually going to see better use. The use cases improve

(42:14) The reduction in costs, the improvement in margins, those things will follow because we have a structured strategy. That’s what I’m seeing the next step, and that’s what leaders need to be thinking about. Right on.

(42:28) The last piece, I guess it sort of brings it up, is this idea that you can’t not do anything here. There’s been a couple other industry, you know, evolutions that say you can’t ignore this. This is happening. And so, if we don’t do anything about this, you know, it’s going to take control of you without, and of your organization. So, as leaders, anybody on the program here today.

(42:54) If you’re in a leadership position, you need to be thinking about this right now. This needs to be an important priority in your organization as to how are you going to control this and direct it moving forward. We have some regular old risks, and I say regular old risks because they’re always out there. But there’s also some evolving new risks. So data leakage and confidentiality is absolutely risk that we would see in any sort of IT or cybersecurity type of environment. But there’s an important piece here that we start to see that emerges in one of them, and I’ll go back to the slide that we had about the respondents to the survey. A number of you all do federal work. Well, the federal contracts mandate certain types of cybersecurity protocols. And when we’re using AI inside of our organizations, we potentially are breaching those contract requirements on on security, on confidentiality, and we could be doing it very inadvertently, and it may not be directly part of the contract, but we have flow-down requirements when there are, you know, federal contracts or federal money involved in anything. And so there’s some compliance issues we need to be thinking about here. We haven’t fully unpacked them, but we recognize those things exist.

(44:05) For example, two-factor authentication, everybody’s using that now, because a lot of contracts and subcontracts require that inside of our own IT systems. We’re probably going to be seeing some similar things of that nature when it comes to AI application. IP loss is another one just to pay attention to our expertise and not letting that slip out the door. Paul, did you want to chime in on that?

(44:26) Yeah, I just sort of sitting here, there’s a couple of old technology maxims, which I think apply to AI just like every other technology that’s ever come down the pike. And the first maxim is spoken like a CIO, my good technology can’t do your good management.

(44:43) Okay, that’s number one. And the other one is technology will take a good process and make it better, faster, and cheaper. It’ll also take a bad process, poorly governed and make it worse much faster. And much more expensive. And that is where we’re at with AI right now, unless there’s some governance around it, it will become unwelcome in certain situations. And it will amplify. It’s a great amplifier.

(45:14) You know, we’ve got to, and I don’t think we need to fearmonger, Eric, I just think that that’s… that’s what the data’s telling us, and that’s the risk that’s… that exists out there right now. So

(45:23) You know, Paul, I would say we’re not crashing the Ferrari by any means, but if we’re sitting here doing donuts in the parking lot and we need to be on a straight track, right? Because we’ve got these powerful performance tools and they’re unguarded right now. And so we can

(45:39) Absolutely.

(45:40) There’s a real novelty to the whole thing right now. And eventually the novelty is going to wear off and then we’re going to be asking then now what?

(45:45) Yeah, yeah. Okay, yep, right on. All right. And don’t forget the Ferraris are rental, so don’t crash it, Eric.

(45:51) It’s a rental, right? We didn’t pick up the coverage by the way. Yeah.

(45:58) Well, don’t they go faster, stop quicker, and corner better? All Red Bulls here. Yeah, and maybe one donut here or there, but don’t crash it. Gentlemen, we made great time and I stand corrected. There was fewer than five slides to get through, so we’ll transition now to our Q&A section. I think we have about 10 minutes. We could probably squeeze in two or three questions.

(46:12) And so just a quick reminder to our audience members, three ways to engage with our speakers. Use the Q&A function in the Zoom panel. You can also share, use the chat box if you have a comment. Our moderators and our team will make sure that comment makes it to our speakers. And last, certainly not least, if you’re willing, just put up your hand using the Zoom functionality in case you want to ask your question directly

(46:33) To Paul and or Eric will give you an opportunity to bring you live into the discussion to collaborate and engage with them directly. But if I may, I’ll tee up our 1st question to both Eric and Paul, and that’ll hopefully give some time for the audience members to reflect on the conversation and tee up any questions they have. And this 1st question is, what is the biggest thing holding organizations back from adopting AI more broadly? Is it security? Is it cost? Is it training or something else? Eric, your slide earlier touched on this, but I’d love to toss to you just for any other further elaboration?

(47:09) I think of that question, it’s not cost. So that’s a non-issue, right? And almost to a detriment. Again, if AI was a lot more expensive, it might make it more strategic. There’s no question about that. That being said is it really is, I think it’s unknown

(47:24) And I think what we saw in the respondents is that people have different concerns with, you know, adoption, and those concerns are broad enough to say that some of those fears could be unfounded, some of them could be rock solid, but it’s largely unknown right now, and I think Right now, our industry confronts us with a lot of use cases, right? We’re seeing everybody pitch something to sell us. Hey, use it to do this and use it to do that. But that unknown, what happens if, and I don’t think we have a lot of authority in the industry talking about addressing those risks, and that’s where I see, I think, the challenges are coming from.

(48:03) Paul, anything you would add to that?

(48:05) No, I think we’ve talked about it. I really think that the the work that’s ahead of us right now is the harder work of putting programs together at a corporate level, which requires the cooperation of a lot of people, some dedicated funding and focus, and some accountability for actual and real and measurable results. To kind of go back to Albert’s equation. And I think organizations need to start figuring that out. So we’re moving out of the phase of novelty and we need to move into the phase of good old-fashioned roll-up-your-sleeves management, which means let’s get the team together, let’s figure out what we want to do as a group, and let’s fund it, let’s measure it, and let’s be accountable for the outcomes.

(48:54) That’s how it’s gonna… that’s really the next phase that needs to happen here, from novelty to real integration. The opportunity is real, but to Eric’s point, we need to… there’s a lot of people selling snake oil, you know, and companies really just need to identify, instead of the fear of missing out, they really just need to do their own homework

(49:18) Yep. So. Thank you, gentlemen. I’ll do a purposeful pause to just screen the room, see if there’s any other comments, questions that I need to surface.

(49:27) Maybe I have a question for the group. Can you hear me? Yeah. Please. Please, we can hear you, Brian. Go for it. Thank you.

(49:34) Yeah, this is for both Paul and Eric. So with all of your background and having this research now, where do you expect AI to change the construction industry the most over the next few years? Is it estimating, scheduling, safety, project management, or or another category that you guys can think about. It’s getting to look in the crystal ball and give us a little bit of your thoughts there.

(49:56) Yeah, Brian. Go ahead, Eric. Yeah, it’s a great question, and that’s… I’m sure everybody’s going, okay, great, what do we… where is it going to change the industry? You brought up some categories of the business. I think it’s going to start blurring the lines between those areas I think between business development into estimating, that’s going to get a little bit more blurry, and I think between estimating into project management and cost controls and accounting, that will also… those lines will not be so discreet. I think that we’ll be able to use AI to… you know, I mean, really, we can almost use it now to do this but to clean up processes that are labor-intensive and that will lead to better results, more consistent results, and less labor-intensive processes. But that is, again, that has to come with discipline, so that’s where I think there’s a significant change on the horizon. As soon as we get past the novelty of it, I think that’s what we’re going to see some substantial change. I do want to comment, though, when it comes to the regular production, this is not going to get us to dig ditches faster, and it’s not going to get us to be able to paint quicker or assemble any quick or anything else like that, because in the end, this is a labor-intensive businesses. The opportunity is cleaning up business processes, those things that allow us to execute the work.

(51:15) Eric, this reminds me of all those years ago when we were doing field automation work. You know, there’s a real sense of Oh, lost my train of thought here. There’s… there’s just a sense of rework in the whole thing. And if I really believe that AI has a unique opportunity. Remember when we talked about everybody in construction does construction so well.

(51:47) But when construction turns into an administrative function, we don’t all do it that well, right? That’s the part of construction that nobody really is that enthusiastic about. That’s not why people are in construction. They’re not in construction to do the proverbial paperwork AI, I think more than any other technology, has an opportunity to streamline that part of the work in ways that we’ve never even imagined because of, not just because of generative AI, but assistive AI, workflow assistance.

(52:19) But also Agentic, where it’s actually designing tasks, executing on those tasks, anticipating work that needs to be done, and doing it, and then machine learning, where it’s actually learning from its own environment and applying improvements to that and it’s going to manage data better, it’s going to reduce rework, it’s going to do the lifting of many, many administrators, and over time, as these technologies mature, this is going to displace job roles in the way that we’ve currently perform them, no differently than the combine displaced laborers in the field, or a backhoe displaced a shovel. Right? So, the hole’s still gonna get dug, the work’s still gonna get done, the promise, I think, is that it can be done more consistently, with more automation, with more delegation in a way that it’s never… with more accurate data

(53:20) And learning in the environment and applying those learnings and becoming a smarter environment. And that’s worth something. So that’s worth paying attention to and going on your own maturity journey. That’s what I’d say to that.

(53:34) I’m feeling really compelled to ask a follow-up question, but I also recognize that we should probably draw to a close to respect everyone’s time. So maybe we can hold that for a part two conversation between Paul and Eric. Before I do draw to a close, I would love to ask Paul and Eric: Final thoughts, executive summary indication for the group with us today. Where do we move forward? Eric, if we may start with you, please.

(53:59) Sure. These are exciting times, right? Just in terms of so awesome, but we’re potentially at the bleeding edge, as we’ve heard that term so many times of something that will shift the industry that will never go back, right? As soon as these significant changes start to to apply, we’ll never see the business won’t be the same. That being said, is there are so many different processes, different places to apply AI and to learn and improve business processes. It’s not a competitive issue. Not yet. It is absolutely just a risk management issue, is not letting it get away from us. The other… and speaking to that, I want to highlight that, you know, AI does a lot of good stuff for us, and there’s a lot of capability, but one thing is AI in any capacity, does not feel risk. It can identify it, but it can’t feel it. And everybody sitting here today knows that in the world of construction, there’s so many decisions that are intuitive in nature. Like bid day. Do you need to cut a couple of percent, cut something off to make sure you get the job. Do you know your competitors, etc. So there’s an art form to this industry that cannot be replaced, and that we have to always keep in perspective. But yeah, this is exciting times. I’ll tell you, it’d be great to sit here a year from now and talk about, you know, additional. But this is evolving in real time, and we’re generating information in real time.

(55:23) Yep. I would second that. AI is technology. It is not human. And any strategy that you might come up with needs to maintain a human in the loop. You cannot just trust this implicitly and just deploy it all over the place without humans being accountable. And so my recommendation to organizations is that we need to start to develop more frameworks for governance. We have traditional frameworks for governance, for technology projects and project selection and project funding and project management and implementation and solution adoption and organizational change management. This is no different.

(55:59) And I really believe the sooner we get on with the management part of this, the better off organizations are going to be. So put your framework for governance together, put your strategy together, build a budget around this, find some use cases, get your pilot going with some measure of seriousness. But above all, assign somebody accountable to this that’s going to report out on it. And is going to be accountable for outcomes because the opportunity is too great for return on investment. The opportunity is also great for risk that’s unmanaged. So this is a double-edged sword, you guys. So go cut with the right edge.

(56:40) Paul, I’d like to kind of just close with this: Lead this on purpose, not because you have to, but because it’s a strategic opportunity. But do it on purpose, not because it’s accidental or you’re being forced into it.

(56:52) Yeah, agreed. Yeah, it can be exciting. Just take it a step at a time. Yep. And give yourself some grace. We’re only… we’re not even 4 years into this. It’s okay, so… but it’s moving quick.

(57:03) Yes, yes it is. All right, thanks everyone.

(57:06) From my team in the ballpark, it’s pretty evident to see why this topic is so, so popular. And so perhaps certainly this is something that we’re going to keep going on and pushing the needle forward on. It was mentioned earlier that this report certainly will be shared out with all attendees on this live event.

(57:22) And we mentioned it earlier, but this is going to be ongoing research, and so I just want to double down on that message because the value that we see today is great. It’s going to be continual value that we’ll keep paying dividends. And so let me take a moment to share a warm thank you, especially to you, Eric. I know that this required countless meetings, emails, and just energy to put your head and Paul’s head together to generate this report on behalf of TAGCXO. So thank you so much for collaborating with us.

(57:52) Yeah. Absolutely. It’s a lot of fun. Great to work with Paul again. Yeah. Yeah, likewise.

(57:56) And of course. And of course, if you are wondering, yes, as you can see on the screen, the meeting recording will be shared out with everybody, all attendees here, and if you would like one-on-one time with Paul and or Eric, we affectionately call this the VIP lesson. So our team is going to be putting up a URL in the chat box right now. It’s quick and simple to register name, email, phone number, and we’ll make sure to coordinate that one-on-one session with our speakers today. On behalf of all of us, thank you so much for taking time out of your no-doubt busy schedule and spending it with Tax CXO and Red Rocks Advisors. Have a great rest of your day.

(58:32) Thanks, everyone. Thank you all. Have a good afternoon. Thanks.