Insight ON AI in Construction? How JE Dunn Brings Citizen Development to the Job Site

Audio transcript:

AI in Construction? How JE Dunn Brings Citizen Development to the Job Site

Jillian Viner (00:02):

Welcome to Insight On. There's an assumption kind of baked into a lot of technology strategy that the best solutions come from the top down, from the IT department, the innovation team, the people whose job it is to think about this stuff. Well, this week we're pushing back on that pretty hard and we have the receipts. Construction superintendent who had never written a line of code built an AI tool that prevents workers from striking live utilities underground. Marketing agencies solved their own hardest data problem and turned it into a product. A university CIO handed his stakeholders the tools and let them build what they actually needed in a single day. You'll hear all three of these stories in our episodes this week and the thread running through all of them is the same. When you put the right tools in the hands of the people closest to the problem, something remarkable tends to happen.

Joseph Schultz (00:56):

We're spending more time on people than on the data and documentation and that's a big win. We can be more connected as a team with what direction are we going in and back as I mentioned originally, engagement in the field is paramount to a safe job site and that's very important to us.

Jillian (01:20):

I'm going to be honest, when I sat down with today's guest, construction was not an industry I expected to be talking about in the context of AI and that's exactly why I think you're going to love this conversation. Welcome to Insight On. I'm Jillian Viner and my guest today is Joseph Schultz. He's the vice president of Mission Critical at JE Dunn Construction, where he oversees data center construction for a single hyperscaler client, a relationship worth over $3 billion a year. Joe's team is doing something that has a lesson for every industry. They've handed the tools to the people closest to the work and told them to solve their own problems. What a field superintendent built to prevent crews from striking live utilities underground is one of the most concrete human AI stories I've heard. And it started with someone who had never written a line of code in his life.

(02:13):

All right, let's hear that story from Joe. Well, Joe, thank you for joining us today. It's lovely to have you.

Joseph (02:20):

Lovely to be here.

Jillian (02:21):

As the daughter of a general contractor and now an office worker who sits at a desk all day, I am particularly excited to talk to you because you are in an industry that I don't think people immediately think about an industry being appropriate for AI, which is construction. So before we get deep into the conversation, just tell me about the company that you work for at GE Dunn and what your role or responsibilities are.

Joseph (02:46):

Very good. Yeah. So I work with JE Dunn Construction. We're a general contractor in the commercial and industrial space. We are around the US, we have 26 offices, top E&R that's a revenue ranking firm of all the general contractors. I personally am focused on data center construction, so VP of mission critical in that group. And my particular role even within that is focusing on one hyperscale data center client. So I'm really the relationship and account lead. And so I see it as my role to make sure that our teams understand our client's values and deliver on that. So we're constantly innovating for them. We're constantly investigating what's important to them and just making sure that we're delivering that in their projects beyond just what's designed.

Jillian (03:43):

So mission critical is the data center part of the construction. Imagine you're just a little bit busy these days.

Joseph (03:50):

It's an interesting time.

Jillian (03:53):

What has the explosion of generative AI done for you in your business besides the data center growth?

Joseph (04:00):

So as AI has been developed, a lot of the discussion around is like what jobs is it replacing for or tasks is it replacing for? And construction always comes up as something that it will not, cannot replace, but it really is a big enhancement in that the human component of construction really is everything. To be able to communicate what's on the drawings and turn that into plans and actions for the crews to then go execute on. So that's really why AI is really an amazing thing to be able to translate the communication quickly into what we need to go do and to what we've done.

Jillian (04:51):

What's been the biggest impact so far with AI communications? How is it helping in that area? Are we talking about email summaries or is it deeper than that?

Joseph (05:01):

It is deeper than that. That's a good one though. I mean, right now we spend so much time just in meetings that our engagement in the field becomes much more challenging and we're looking at our teams to be highly engaged on the job site, boots on the ground because when we're engaged, we win from a safety standpoint. We are better at protecting the risks, keeping people out of harm's way and making sure everybody goes home well. We can do that when we're highly engaged. When you see onsite engagement drop, that's when you start to see these ankle biters per se come up from a safety standpoint. So for one, just on that, the email summaries or meeting minutes or things like that allows us to spend more time in the field because we're spending less time in meetings or I'm able to cover for you better because we can communicate you sooner.

(05:59):

So that's an easy win right there. But the other ways that we're using it is that it can connect data sources in a much more streamlined way where back to the human thing, what we're doing is we're trying to get people together to look at a thing we're about to go do. We're about to go dig in the ground, for example. And so now through these tools, we have a better way to get all the plans together of what might be in the ground already when we need to go do this task aligning on a time to get everybody together and have a discussion around that feature of work and say, "Are we clear to go do this? " It's been really amazing that when we use that process we don't run into problems and when we don't use the process, sometimes you will. You'll have a waterline strike, a fiber line strike or something like

Jillian (07:02):

That. Because the AI has like all the context and can-

Joseph (07:06):

It's able to very quickly pull together all that information.

Jillian (07:10):

It basically is finding new blind spots for you before they happen. Interesting. Let's be real for a moment because I think a lot of people are exhausted a litle bit by the hype of AI and what it supposedly can do. I myself am trying to use AI to redecorate a room, which you would think it'd be easier to take a picture of it. I say, "Hey, put this desk in here, change the wall colors." And it can't remember that the window's on that wall and the bed's on that wall. What are some of the limitations that you've seen in the construction space?

Joseph (07:41):

What's tough for us is I would phrase it as grounding. We have requirements to deliver on, be that the drawings of what the physical space is supposed to look like at the end of the day. The specifications of these are the requirements to go do the task or these are the materials that need to be used or these are the quality checks that need to be done. There's requirements to deliver on. And so that's what I'm really excited about now is there's been a really good development in grounding AI on a rule set that cannot change or a rule set that needs to be true and then using its power to really find the solution for it. So back to communication, it's like our projects are built on specification documents that are this fat, black and white text and you think that's a rich document to really look at how you're going to go do this.

(08:44):

AI is a really fast way to be able to, with the scope I need to go perform, find everything related to it and in a way that I can digest it.

Jillian (08:54):

Synthesize the information. Super helpful. Let's talk about this concept of citizen developers. Because again, I think when you think construction, AI doesn't necessarily come to mind immediately, but this is actually something that you guys have been able to embrace. What has that looked like?

Joseph (09:13):

It's been a fun journey and we're investing more and more in it every day. So Citizen Development, this is where we can design agents, applications, skills, things like that for us by us. So I'm an operator, I'm a project manager and I know best what a project management's job needs and I know best what a project management's bottlenecks are. Same for our field supervisors, our superintendents, our safety professionals and everything like that. So we've got a team in our IT department that is focused on us to develop our own tools with their guidance of how we should go about it. So we've got a team of folks focused on citizen development. They spend quite a bit of time on our job sites really just seeing what takes us time to go do our job.

Jillian (10:11):

What kind of stuff is that uncovered?

Joseph (10:14):

So one I kind of mentioned a little bit. Before you go do a risky task, you have to essentially file a permit to go do it to use a tool that maybe has more risk than another. You've got to go file a permit that would say, "I know how to use this properly. I'm going to use it for this task and then I'm not going to ... " Same for like, it's funny, even a ladder, you need to file a permit to use a ladder because they are riskier than a mobile work platform.

Jillian (10:47):

Okay. So there's a lot of paperwork involved.

Joseph (10:50):

A lot of paperwork, a lot of checks and balance and making sure that you've really thought through the risks set in hand before you go do this task. Well, so one of those, the big one is the excavation permit requests. We call it a dig permit. And so like I mentioned, the whole point is to not strike any utilities if you're, especially on a brownfield site, we're working on data centers. It's imperative for their uptime that we're extremely careful and aware of where we're digging in the ground. So this has been a phenomenal way through an application developed by a superintendent to eliminate risk in these activities.

Jillian (11:34):

Tell me more. What did that actually look like? How did they go about this? What does the solution do?

Joseph (11:40):

It was taking a fairly archaic process of just saying, "Okay, we see this work coming up on our schedule. We need to go out there and walk the space and people are going to bring their drawings." And usually some people wouldn't show up, some people would show up unprepared. And so too often this meeting was happening without the right information. And so are we ready to go do this task? Well, we're not sure because we didn't have this vendor here with the right information and so forth. So this was a governed process that's very slick to use to capture all the information and the people so that we know when everybody shows up, we've got the right stuff, it's been overlaid and we're ready to go.

Jillian (12:27):

That seems like a very transferable lesson because the before that you just described, all I can hear is delays. That's

Joseph (12:34):

What that

Jillian (12:34):

Causes. Delays, risk, uncertainty.

Joseph (12:38):

It's delays or risk or both.

Jillian (12:41):

Or both. And this is something that has to get done almost repeatedly. Every time you go to a different job site, you have to go through this exercise. And so you've identified a common pain point and someone who's close to that work said, "Aha, I think we can make this better. I think AI can help make this better." And that's what they did. They basically created this process with AI. Brilliant.

Joseph (13:06):

And other ones in safety that we're working on now is before any crew goes about their day, they have to do a job safety analysis. And so they're going to look at, okay, we're going to install these doorframes today, let's just say. So what are the risks in installing doorframes? Do we have sharp edges we're going to be working with? What about the areas we need to travel through? Is there risky work going on there? What kind of tools are we using and risks inherently with that? And so they go through a process of as a crew discussing what they're going to do for the day and then filling out on a form that kind of guides them through the process of thinking of everything, but it's too easy to just check the box on that activity. And so very excited with AI of now just letting a phone start recording the conversation happening and providing us feedback of, is this a rich conversation about risk and have we covered all the bases truly so that we know they're not checking the box and we know they really have discussed it

Jillian (14:13):

Interesting. Safety really is at the core of those types of use cases to make sure that the humans are doing the work that really needs to get done. How else are you using AI on the job site? Because again, that seems like an odd place to think about AI.

Joseph (14:31):

Looking at it for some of our accounting tasks, I don't think we're yet willing to trust it with math yet we're seeing-

Jillian (14:41):

One of those limitations-

Joseph (14:42):

One of those limitations, but some things that we saw today is really around projects and creating agents that we can manage and govern within projects. So that comes back to the grounding of if I could, rather than just having a folder structure of the documents, have let's say a Gemini project with all of the grounding information put in that of plans and spec, safety requirements, contract requirements, so that really the team can work in that ecosystem and be poking at it all the time with question about this, answer about that, to find the best solution for nearly any problem we encounter.

(15:36):

I don't know if it could replace the lived experience, but it could give you really good guidance of saying, "Hey, I'm having a delay here because of this. How should I manage that delay and make sure I'm meeting all the client's expectations?" Well, that could take you a long time in going through the contract and trying to interpret legalese when AI can really spit out you in very plain language, all the bullet points you need to do and if you go and track through those bullet points and execute on those, you've met all the client expectations and then some.

Jillian (16:10):

Looking back to a year ago, how is your job or your work environment different now with AI than it was before?

Joseph (16:18):

We're spending more time on people than on the data and documentation and that's a big win.

Jillian (16:30):

What does that mean for your people? What do they do with that time back?

Joseph (16:34):

It means we can be more connected as a team with what direction are we going in and back as I mentioned originally, engagement in the field is paramount to a safe job site and that's very important to us. So if we can through cleaner communication, more to the point, quicker documentation that we can broadly share out, if we can cut time out of people's meeting calendars, that's time they can spend in the field.

Jillian (17:06):

Yeah. Because you've gone through the experience of citizen development, people on the ground identifying problems, what's the biggest lesson or takeaway that you have from that that you would advise other leaders who are looking for ways to use AI, particularly in fields where it's very physical, very hands-on? What can they learn from this?

Joseph (17:28):

Starting is fairly simple. You see these amazing use cases out there that could be very complex and you could spend a lot of time brainstorming of how to get started there, but really to just start poking at it, let your team experiment with it, encourage them to do so and really they're going to start coming to you with great ideas. And that's back to our citizen development is distributing the amount of creativity and thought is a force multiplier with that rather than we're going to have just these three folks be our task force with AI. They're going to develop the tools so let them do it. You really need to distribute that work and bring more concepts into it through just more brains at it of the creative thing.

Jillian (18:17):

Any fears of risk or it's really just more about coming up with the best solutions and you see that that happens at the front lines of work?

Joseph (18:26):

I'm really impressed with everybody so far using it for the best intent possible and thinking intentionally about if we truly do have a requirement to deliver on or a standard we need to be above, so far I've seen like that's all been in mind when developing something and that if a solution provided by AI maybe seems too experimental or untrustworthy, it's taken as such of, "Okay, we got somewhere here, but I'm not going to take this and run with it, but okay, we can work on this. " So everybody's been very mindful of that. That's been less scary than I think I would've felt two years ago.

Jillian (19:15):

What about the future? Where do you think AI is going to change things in the future? What do you hope it can bring to your work in the future?

Joseph (19:21):

I look to the same of that it really enhances the human interaction. That's what's extremely important in construction and just for humanity in general, that's I think where we'd all prefer it to go is just that it does not replace, but it enhances our human interaction. I think about how we onboard people to our company and our projects. So construction's a unique industry in that we need to create essentially an organization of a couple thousand people the size of these job sites. We create an organization of a couple thousand people, all of whom are temporary, right? Yeah, they work for their respective companies, but if this project is a company for two years, let's say, everybody's temporary. And so to think that people are going to show up and know what to do, know our standards of operating, know our culture of working with each other, know the ... If we have requirements above the industry, knowing that those requirements are above the industry and how to abide by those, to think that everybody's going to understand that well is a big ask.

(20:44):

And so that's something I'm very excited about from AI, from a communication standpoint, is you show up to a new project and you're presented with a pile of information and it's like, "Here you go. " Well, we can find much more dynamic ways to digest that information, much more interactive, much more engaging. And so if that gets you on board faster, that's a huge win.

Jillian (21:10):

Sounds like a big win.

Joseph (21:12):

Yeah.

Jillian (21:12):

Joe, thank you so much for the time today. I really appreciate it.

Joseph (21:15):

Thanks for having me.

Speaker 3 (21:16):

Thanks for listening to this episode of Insight On. If today's conversation sparked an idea or raised a challenge you're facing, head to insight.com. You'll find the resources, case studies, and real world solutions to help you lead with clarity. If you found this episode to be helpful, be sure to follow Insight on, leave a review and share it with a colleague. It's how we grow the conversation and help more leaders make better tech decisions. Discover more at insight.com. The views and opinions expressed in this podcast are of those of the hosts and the guests and do not necessarily reflect on the official policy or position of Insight or its affiliates. This content is for informational purposes only, should not be considered as professional or legal advice.