AI in Construction: Powerful Tool or Wasted Opportunity?

🔗 LINKS

Visit the episode page for more details and a transcript of the show.

Sign up for group coaching to get your Cash Flow MRI

Get Profit First for Commercial Construction--OUT NOW!

Subscribe to get notified as soon as new episodes go live.

ℹ ABOUT THIS EPISODE

Are you using AI like a $500K employee to check the weather? Wade Carpenter and Stephen Brown break down how construction contractors can stop treating AI like a search engine and start using it as a strategic business tool. 

xFrom punch lists and field reports to change order negotiations and equipment decisions, you'll discover practical, cash-flow-focused ways AI can sharpen your thinking — without replacing your judgment.

⌚️ Key moments in this episode:

• 00:00 Stop Using AI Like Google
• 00:29 Show Intro and Hosts
• 02:42 AI as Your Employee
• 04:18 Real Contractor Use Cases
• 06:07 Voice Notes to Punch Lists
• 08:15 Ask Better Questions
• 10:28 AI as a Panel of Experts
• 11:34 ROI Is Better Thinking
• 12:54 Context Makes AI Smarter
• 15:23 Don’t Delegate Judgment
• 16:57 GPS and Tesla Flood Story
• 18:39 Final Takeaways and Wrap Up

The Contractor Profit Blueprint is a complete guide that breaks down exactly how to identify where your money's going and start keeping more of it. This isn't theory. It's the same framework I use with contractors I work with every single day.

Head to profitfirstconstruction.com/blueprint to download your free copy. 

Stop working for free. Let's get you keeping what you've earned.

Join the Profit First for Construction community!

Find all episodes and related links at ContractorSuccessForum.com.

Join the conversation on our LinkedIn page: https://www.linkedin.com/company/CarpenterCPAs
   
FIND US ONLINE
Wade Carpenter, CPA, CGMA | CarpenterCPAs.com
Stephen Brown, Bonding Expert | SuretyAnswers.com

Transcript

[00:00:00] 

Wade Carpenter: What if I told you, you hired a half-million-dollar employee last month, and you've been using them to [00:00:05] answer questions like, "What's the weather tomorrow?" Sounds ridiculous, doesn't it?

But that's [00:00:10] exactly the way most people are using AI. It's like buying a [00:00:15] Ferrari and using it to drive to the mailbox. Sure, it'll get the job done, but you're completely missing what it [00:00:20] was built for.

Today, I want to show you why using AI like Google is [00:00:25] costing you opportunities, time, and maybe even your competitive advantage.

This is the [00:00:30] Contractor Success Forum. I'm Wade Carpenter with Carpenter Company CPAs, alongside Stephen Brown with [00:00:35] McDaniel-Whitley Bonding & Insurance. Stephen, do you think our listeners are getting all [00:00:40] they can out of AI?

Stephen Brown: I most certainly do. I'm kidding, Wade..

Wade Carpenter: [00:00:45] Okay. It looks like this is gonna be a short episode then.

Stephen Brown: No. I love [00:00:50] your content. I use AI every day. Everybody in our office uses AI. We [00:00:55] subscribe to it. We have a closed AI system. That's important to [00:01:00] us. And how you use it and what you use it for and what you trust it to do is everything.[00:01:05] 

And I love your analogy about hiring a half a million dollar employee and asking him, [00:01:10] "How's the weather?" How do you get to your mailbox? Okay.

So yeah, you're right. The Google [00:01:15] AI feature is a lot better than having no AI search engine at all. [00:01:20] It does get to the point very quickly on things, and we're not gonna get into that [00:01:25] so much as the whole mindset of why and how you're using AI to help your [00:01:30] business.

Wade Carpenter: Yeah, and this is not about, the Google chat engine and all that stuff. Even if you're using [00:01:35] ChatGPT, if all you're asking-- A lot of people are out there watching these YouTube videos [00:01:40] and seeing things like Claude Code and saying, "I just don't get it." like I said, I've learned a lot of [00:01:45] that kind of stuff, but I still get questions like very basic things.

So many of our [00:01:50] contractors are out there asking ChatGPT, "What time does Lowe's close?" How [00:01:55] does that help anybody? Yes, maybe that's a little easier than what you're talking about getting [00:02:00] through Google AI, but what I'm hoping to do is start opening up that productivity, [00:02:05] even if you don't want to go into Claude Code and those kind of things.

But before [00:02:10] we get into that, if I could, a quick word before we get into the points [00:02:15] today. If you get one usable idea out of the next 20 minutes or so, I [00:02:20] would really like you to do two things for me.

One, if you would hit subscribe. That is [00:02:25] what puts this show in front of other contractors who really need it.

And two, [00:02:30] if you would, share it with somebody, your project manager, your estimator, your office manager. [00:02:35] Honestly, especially your office manager, I guess I would say. Half of what we're talking about lands [00:02:40] on their desk before it lands on yours.

For a lot of people, it's just how do we [00:02:45] get beyond using it as just a search engine?

Point number one, [00:02:50] getting beyond AI as a search engine. It can be an employee. So like we [00:02:55] said, nobody really hires a project manager to answer a trivial question. You [00:03:00] hire them to organize your chaos.

Organization is one of AI's [00:03:05] superpowers, and all Google does is retrieve information. AI can [00:03:10] actually organize that information, so there's a huge difference in that.

Some of the things we were talking about before, [00:03:15] Google can send you a bunch of pages and you're searching through a bunch of paid ads and stuff like that.

[00:03:20] But AI can give you perspective on the topic, and that's what one of the easy [00:03:25] unlocks we've got here. You may be asking the same question whether it's ChatGPT or [00:03:30] whatever, you're not just searching, you're collaborating, and you can ask questions and follow-up questions [00:03:35] and how does this affect me?

Does that make sense?

Stephen Brown: It does. Exactly what you're talking about. [00:03:40] So many instances that I want to share about how some of my contractors and [00:03:45] how we at McDaniel-Whitley in our office are using AI to really help us [00:03:50] get to the crux of the matter quickly and to summarize things, to [00:03:55] help us organize things and how to move forward.

You can imagine in my [00:04:00] business there's nothing but insurance policies, and a lot of them are different from our different carriers. There's [00:04:05] different endorsements, exclusions, and also they're contracts our customers are signing [00:04:10] that they have to be covered for and they have to know what their risks are.

Thanks, Wade. What [00:04:15] practical ideas did you want to share with everyone about this? 

Wade Carpenter: It sounds like you got some ideas, but let me [00:04:20] throw one example out there. We talk about replacing your lawyer and stuff like that. But, one of the [00:04:25] questions I get is, what is my labor burden? How much would it cost me to hire another [00:04:30] employee?

And, you can ask, whether you got ChatGPT, Codex, or Claude, whichever one, [00:04:35] just, say, "Okay, ask me questions. How do I figure this answer, and what's [00:04:40] it gonna take for me to be able to afford another project manager or something like that, or an [00:04:45] estimator?" And so it can ask " what's your crew like and what's your volume and can we cash flow this [00:04:50] kind of thing?"

But, instead of Google throwing out 40 different articles of [00:04:55] people that wrote a bunch of things, maybe there's some good points in there, but you have no idea. [00:05:00] Ask a question like that. You're asking the same question, but you get a whole different answer for that.

[00:05:05] What you were saying is summarization and organizing chaos. When you don't know how to put it [00:05:10] together in your head, that's one example I had.

Stephen Brown: Another one, Wade. I don't know if you agree with me, but [00:05:15] sometimes you'll ask it a question of how to do something, and it'll give you too much information, and then [00:05:20] I'll ask it "Summarize that and give me the best points." You'll read a whole [00:05:25] section and something will make sense or really hit home to you, and you want to dial in on it [00:05:30] further, and then you can ask it to help you do that for you.

I guess the whole [00:05:35] conception of having it design a webpage for you, having it design [00:05:40] functioning software for you to get certain projects done.

This part of a tool is just [00:05:45] boggling most people's brains. They're not seeing what it can do. I [00:05:50] think of AI as the difference between a hammer and a nail gun. Seriously. What a huge [00:05:55] advance. And that's just it. It's just a tool. It's a computerized tool, but it's a [00:06:00] tool.

What are you seeing everyone using AI for right now, Wade, that [00:06:05] is really you want our listeners to know about?

Wade Carpenter: We'll sort of get into some of these other ideas in a [00:06:10] minute, but, just using AI instead of if you're not a great typist, just dictating [00:06:15] whatever the idea is. I do that all the time. But think about, trying to walk through like, your [00:06:20] punch list or something.

Say you're doing a walkthrough and just dictate like, " turn this into [00:06:25] a punch list." "This is what I see on this room," turn it into a punch list and organize it.

[00:06:30] Maybe you're the general contractor. This is a little electrical problem. This is a plumbing problem. Just [00:06:35] say, "Turn this into a punch list for who's responsible for it," and flag [00:06:40] anything that sounds like a change order. Maybe give it to the contractor.

Stephen Brown: No, that's a great idea. [00:06:45] And take it even further, Wade. You do that, say you have a voice [00:06:50] recorder, it's small that sticks to the back of your phone, and you push a button and you record something. Then [00:06:55] you take that digitalized voice recording and you download it into AI, and you [00:07:00] have it summarized, and you have it do that punch list for you.

And then what do you do [00:07:05] with it? That's as important of how you use an AI, what you do with it, how you organize it, [00:07:10] how you don't forget you've done it.

Wade Carpenter: And I didn't want to get into specific tools, I [00:07:15] have subscriptions to all of them, but ChatGPT recently came out with their [00:07:20] voice mode, which, Claude did too, but ChatGPT Codex, that [00:07:25] sounds like you're talking to a real human.

If you want to get job report, field reports from your people every [00:07:30] day, and your employee's got ChatGPT on their phone when they're driving home, [00:07:35] just have a question like, "What was the weather like today? Did Did we get slowed down by something?" And it [00:07:40] can actually ask you questions and then turn that into a report and turn that into [00:07:45] management. And I don't know if you tried it, but it's actually like talking to a real human. 

Stephen Brown: Yeah. I [00:07:50] do it. And, to me, sometimes I get contacted or I talk to someone [00:07:55] and they identify themself as a virtual AI assistant. 

Wade Carpenter: Yeah, we don't want to do that.

Stephen Brown: I know, that [00:08:00] kinda ticks me off because I'm like, "Really?" And then they do the [00:08:05] "Yes." "Oh, yeah. I'm sorry, Stephen. Yeah, I'm a virtual AI assistant," and [00:08:10] you're right. I know. Okay. So please don't let me distract you.

Wade Carpenter: Well, it's [00:08:15] okay. The second point which, first of all, stop looking for answers and start looking for [00:08:20] better questions. That was point number two. 

Hear me out here, this is one of the biggest shifts I noticed AI [00:08:25] often gives me better questions than answers. That's where the value is.

It's like, okay, [00:08:30] this is my situation. Say, "Should I buy this piece of equipment?" Instead of doing that, [00:08:35] say, what questions should I be asking before I buy this piece of equipment?" Suddenly [00:08:40] AI can become your checklist generator. Those kind of things that you can't-- assessing a [00:08:45] risk, a coach.

Stephen Brown: Absolutely. You're right. It is really good at asking [00:08:50] questions and pushing how your thought process might need to change a little [00:08:55] bit.

Wade Carpenter: The classic one we talk, like, what should I ask before I buy this [00:09:00] $180,000 excavator? Thinking about what is your utilization rate? What's it gonna do to your cash [00:09:05] flow? What happens if it's sitting idle in January, February, March? How much more [00:09:10] revenue do you need to hit?

Again, lot of these sort of go into the legal and tax [00:09:15] sides and, that's not a replacement for your lawyer or your CPA. But I mean, you know, [00:09:20] asking about like, Section 179, bonus depreciation, who maintains it? What am I [00:09:25] not thinking about, like maintaining it, insurance on it?

Stephen Brown: Yeah. Ask it the most simple [00:09:30] question and then dial down on it. It'll give you more than you want. Don't be intimidated by [00:09:35] that. Scroll through it, see if it's pointed you in the right direction of what you need to [00:09:40] study some more and you need action on, and then rephrase that.

Wade, as an [00:09:45] aside, a lot of times AI will run a list of ideas, and at the bottom it'll [00:09:50] say "We could further expand this by blank, and blank." And you say that one [00:09:55] looks good." And I go to copy and highlight it and paste it down below to [00:10:00] continue, and it won't let you do that because you can't copy and paste off the AI platform, [00:10:05] so you have to retype it in there.

Is that what you run into too, or-- 

Wade Carpenter: Yeah, all [00:10:10] kind of stuff like that. But, asking it, " Should I bid this job?" Or, it's [00:10:15] a lower margin, it's a bigger dollar, just say, "Give me 10 questions I need answered [00:10:20] before I should put a number to this thing." Just using it as your thought partner. 

I want to get [00:10:25] to a lot of these points too, and we may have to skim through some of these.

But, point number three that I [00:10:30] wanted to make was more about AI's sort of a room full of specialists.

Just [00:10:35] saying, give me one opinion, tell it to act as my attorney or, act as my [00:10:40] client. How would my client see this? Act as, like, how would your superintendent see this that [00:10:45] doesn't know what's going on in the office? How would your toughest competitor see you? You [00:10:50] know, those kind of things.

When you tell it to give it a persona it gives you different [00:10:55] viewpoints. I don't want to get into like setting up agents and stuff like that, but when you give it different viewpoints, you get [00:11:00] different answers and different ways that we're not maybe thinking about. We get this tunnel [00:11:05] vision.

I wanna give just a couple of examples and make sure we keep moving through this one. I was negotiating a [00:11:10] lease, and I was trying to see it from their perspective and how to respond.

But let's just [00:11:15] apply this to a contractor. Let's just say it was a change order fight. And we're going through back and forth and [00:11:20] everything. Like, Okay, you're the project manager for the GC. How do they think about it? How [00:11:25] does the owner think about it? You're my attorney. How would they think about it?

Acting as your CFO, [00:11:30] how do we frame this like, okay, this was not in the scope. Just another example.

The next point I was [00:11:35] gonna make, the ROI of this isn't just speed. Everybody talked about saving time [00:11:40] with AI, and we wanna do that, but that's underselling it the way I see it.

The biggest return is not [00:11:45] typing faster, it's thinking better.

And so I feel like contractors lose money [00:11:50] because they don't lose money because they type slow. They lose money because they missed something, [00:11:55] they forgot something, they assumed something. They didn't ask the right questions. [00:12:00] And so an extension of the last point is AI is helping expose those blind [00:12:05] spots.

Stephen Brown: You might say an insurance agent or a bonding agent is a middleman. They're in [00:12:10] between me and the companies that I do business with. That could be a problem if [00:12:15] you're slowing things down.

We had a situation where just simple meeting [00:12:20] where a contractor was selling us on a certain project and why [00:12:25] the bonding company needed to consider it. And with his permission, it was [00:12:30] recorded. It was absolutely perfectly communicated by the [00:12:35] underwriter. We took it, put it in AI, condensed it, sent the summary to the [00:12:40] underwriter with our recommendation, and it was approved.

Not [00:12:45] only was it approved with thanks, thank you, yes, we approve, because [00:12:50] everything's documented.

And that's the way it is in my business. I know it is in yours too.

Wade Carpenter: The [00:12:55] next point is one thing I guess I would say I've discovered, AI is almost like a mirror. It [00:13:00] reflects the quality of your thinking. And so if you give a bad prompt to it, [00:13:05] you're gonna get a bad answer.

It's sort of extending that blind spot. If you [00:13:10] give it great context, you feed it all this information. You were just talking about recording the conversation, turning it [00:13:15] into speech using something like Otter.ai or all these speech-to-text things.

[00:13:20] It's easy to do now. But if you got all that context, you can turn it [00:13:25] into an incredibly talented intern, I guess I would say. if you give it vague instructions, you're gonna [00:13:30] get a bad answer. Just think about the scope of it and are you [00:13:35] asking the right questions?

Maybe that's a question you should be asking AI, is "If you were my [00:13:40] consultant, what questions should I be asking?" 

Stephen Brown: Yeah, I like that analogy. Now, [00:13:45] I heard the other day someone say, "You can't look at yourself in the mirror and lie to [00:13:50] yourself." So if you're trying to work through something, just talk to yourself in the mirror. [00:13:55] Thinking of AI as a mirror of reflecting your thought process is one thing, [00:14:00] but also using that to fine-tune your thinking.

There's emotions involved. [00:14:05] Every time you make a decision, there are emotions involved. You take a few minutes and you help [00:14:10] AI sort through those emotions, that's just another great tool. 

Wade Carpenter: You really wouldn't hand that brand new [00:14:15] PM a $2 million job that you've never really worked with and just say, "Go handle it," [00:14:20] and then act surprised when things happen. It's the same principle. You gotta give them good [00:14:25] context and stay on top of things.

AI is only as good as the context you [00:14:30] give it. A lot of people learn ChatGPT, and they're really no farther along than [00:14:35] they were November '22. They quit too early. They ask one question and read [00:14:40] one response, close the window and be done with it.

And so maybe the better [00:14:45] approach would be like, you ask a question, ask a follow-up question. Keep asking, "Tell me more about [00:14:50] this. What am I missing? Give me three alternatives." Challenge your own answer.

What's [00:14:55] amazing is I'll take something in Claude or maybe ChatGPT generated, throw it [00:15:00] in Claude, "What's your perspective on this?" And then I'll also like Grok or Cursor, I'll [00:15:05] throw it in there.

It's amazing also if you can get these agents to disagree with each other. It's [00:15:10] w- the kind of I won't get too geeky on here. I'm not trying to do it, but the point of it is, [00:15:15] we quit too early. So asking, "What am I missing?" Take the [00:15:20] opposite position. It's amazing what you get out of it.

The last thing which I [00:15:25] tell my employees, that's what separates a good accountant [00:15:30] from a bad accountant. There are people out there just throwing stuff in AI [00:15:35] and just taking their answer as gospel.

That human element you were talking about a while [00:15:40] ago where, it's like, "Oh, I'm talking to AI." It's like canned answer. You can't delegate the [00:15:45] judgment. That's the biggest part.

Again, replacing your attorney, it can go a [00:15:50] long way for, maybe starting some of the questions, but, this is where people, they [00:15:55] rely too much on it.

You can't just blindly take your AI and make [00:16:00] your decisions based on what AI tells you either. A lot of times, we were talking about the context, they don't have [00:16:05] that. If you just blindly say, "This is the answer," and it's like, okay this is what their [00:16:10] project manager should have told me, or, my accountant or my attorney.

They all should have input, [00:16:15] but, you as the owner, if you're the owner listening to this, you own that decision [00:16:20] and you've got to make it for yourself. And so don't just rely on [00:16:25] whether it's a human or AI.

Stephen Brown: Yeah, no, you're exactly right. You should use AI to [00:16:30] fine-tune your thought process of what you want to talk to your attorney about. [00:16:35] Absolutely. I'm just using the attorney for an example. You can't get rid of a seasoned [00:16:40] construction attorney with their knowledge, their experience, their local knowledge their [00:16:45] intuition, but it can help you define the things that you need to be concerned about. And [00:16:50] of course, your attorney is doing it as well.​[00:16:55] 

Wade Carpenter: I looked at my glove box the other day and I pulled out [00:17:00] an old map. It's like well, you remember those things? You know, And like I said, I think [00:17:05] about GPS out there. Now we rely on that. But, just like AI your [00:17:10] GPS can suggest a route, but it doesn't know that there's a tree in the road. Maybe now it can, [00:17:15] but you know, like the traffic, there's a wreck or something like that. Or the road washed [00:17:20] out, something like that. I don't know. The driver needs to look through the [00:17:25] windshield, and AI is just sort of like giving you routes. But if you only pay attention to that one route, that's [00:17:30] the point. So don't delegate the judgment.

We've talked about it before. You got a [00:17:35] superintendent going through a divorce or, maybe you're struggling with an owner that [00:17:40] pays 75 days late, but you've worked with them for a long time.

They've never stiffed anybody in 20 [00:17:45] years. I don't know. Just thinking about, like your banker. maybe they'll tolerate one bad [00:17:50] quarter but not two. You know, It's just thinking about the direction.

Stephen Brown: So were you talking about consistency [00:17:55] there or where were you going with that idea?

Wade Carpenter: The idea is we can't delegate [00:18:00] the judgment. Don't delegate the judgment to AI is what I'm saying. [00:18:05] Make your own decisions. That's the point. If you just go down a road that flooded [00:18:10] last night, it may spit out an answer effortlessly, but it may not give you the right [00:18:15] idea with all the context. That make sense? 

Stephen Brown: It's funny. I was telling you just [00:18:20] the other day about my Tesla running me right to a flooded road, and I thought it'd sweep me down [00:18:25] into the river. The water was up halfway over my windshield, and it just trudged right through, [00:18:30] but it was following its directions.

Wade Carpenter: Well, That's a excellent point. [00:18:35] There you go. Yeah, I remember you telling me that. [00:18:40] But like I said, I don't know if you have any takeaways from this, but the way I see [00:18:45] this, the winners are not gonna be the ones with the newest tools, they're the ones that are gonna get better at using [00:18:50] them.

And so if you're just using it as a search [00:18:55] engine, you know, get a kid to show you. It's amazing, like what somebody that [00:19:00] just, spend an afternoon with some people. I've done that, people don't understand is spend [00:19:05] a couple of hours with people and then just unlock,

Stephen Brown: Yeah. Your comment about get a [00:19:10] kid to help you sort through it is hilarious 'cause that's exactly what I do, and we do at [00:19:15] McDaniel-Whitley. Our kids come in and go, "Really? You doing it that way?" And we're like, [00:19:20] "Yeah. Let me show you something." And you know they know a lot.

They certainly know a lot [00:19:25] about that particular technology and what it's doing for them. But the things they don't know [00:19:30] at their age are incredible as well. So I think that's a great way to tie this all [00:19:35] together. 

Wade Carpenter: I know a lot of our listeners, they have a lot of questions. I've actually been asked to teach some more [00:19:40] of this stuff in AI by some people, and we need to relate it to the contractors. I could [00:19:45] get out here and teach Claude code and stuff like that all day, but how does this [00:19:50] relate to you? And that's what I was hoping to accomplish with this.

So if it at least got you thinking a [00:19:55] little different ways then hopefully the time you spent with us was well worth it.

We appreciate you being [00:20:00] here. We do this every single week, and if you haven't already, please share, subscribe.

If you've got [00:20:05] some comments about the ideas or whatever, put them in the comments below. We always [00:20:10] appreciate it. If you got questions, glad to answer them if we can.

Thank you for being here. We do this every [00:20:15] single week, and we will see you on the next show.