AI Can't Fix Chaos: Cut Through the Noise with These Tips

🔗 LINKS

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

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 constantly chasing the next AI tool hoping it will magically fix your construction business? Wade and Stephen reveal why the real shortcut isn't smarter software - it's getting your systems and people to speak the same language. 

Learn how to eliminate translation friction between estimators, project managers, CPAs, and bonding agents before you invest in more technology. Discover practical strategies to streamline your business operations and set the foundation for AI success.

⌚️ Key moments in this episode:

• 00:15 The Real Shortcut: Unified Communication
• 00:56 AI in Practice: Real-World Examples
• 01:16 The Evolution of AI: From ChatGPT to Today
• 01:38 Challenges in AI Adoption
• 03:39 The Language Barrier in Construction
• 05:25 Streamlining Systems with AI
• 09:05 Standardizing Work In Progress (WIP)
• 12:18 The Future of AI in Business

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: Are you constantly chasing the next big AI drop the tool that's finally gonna fix your world? The cheat code that makes your business run itself while you sleep? Yeah... about that. The real shortcut isn't smarter software. It's getting your systems and your people to actually speak the same language.

This is the Contractor Success Forum. I'm Wade Carpenter with Carpenter Company CPAs, alongside Stephen Brown with McDaniel Whitley Bonding and Insurance. Stephen, thoughts on what we hope AI is gonna do for us. What are your thoughts on today? 

Stephen Brown: I don't know. I kind of like your comment, Wade today about, speaking the same language. You know, you remember that book. Men Are From Mars, women are from Venus? I mean, the whole pretext was, men speak a completely different language than women, so you've got to realize that if you want to understand the differences.

Well, you know, the same with AI. People keep expecting AI to do these things for you, and then how does that affect communication? That's huge. I think right now to a woman I was talking to yesterday that [00:01:00] was saying she uses AI to do transcripts and synopsis of her case studies. She's a counselor. And how it's a game changer for her and how wonderful it is. So Wade, I'm fascinated to see where you're going with this topic today.

Wade Carpenter: There was a couple of things that sort of led to this one. I got to thinking, it's November of '25. It was like three years ago, November '22, that ChatGPT really started taking off.

We've gone from, just type something in and it'll spit out some words, initially it wasn't doing a whole lot of anything. It would write some stuff for you. And it was probably better than us trying to do it ourselves, but it still didn't do things like math and things like that very well.

But I was having a conversation with a contractor this past week, and it was obvious to me that-- I spent a lot of time learning AI. He's done the same thing, but he's also pinning his hopes on, hey, we're gonna be a lot more efficient. He's gonna buy this CRM that does all this stuff. And I gotta thinking, okay, is that really gonna make any difference in there?

I mean, we all can [00:02:00] see in three years, AI has made some huge leaps forward, and we can talk about some of that. But I'm just wanted to explore today, maybe this is just a thought experiment of where are we going and how do we make it really work for us? What is that cheat code that we can plug in and start to really see some real advantages to it? 

Stephen Brown: That's a great question, I think. Because, I mean, I'm kind of following you what you're saying about the cheat code, and I'm thinking of all the things that we're using AI right now for in my business. You know, you think about so many different things that you would easily become overdependent on AI doing for you. Just like adding something on your calculator, you adding a bunch of big numbers and you're trusting your calculator to provide the exact right answers. And this just could be millions of dollars. It doesn't matter, but you're trusting your calculator, right? You used it for so long, but. on this, AI, I think to myself, what about putting a full contract in there, a construction contract in there, and depending on AI to find the things that you want it to [00:03:00] find?

What if it doesn't find it? Who are you gonna sue, Chat GPT? Who's responsible for that? Ultimately you are, right? It's just like if you don't know how to add and subtract, multiply, and divide, then you can't sue the calculator company.

Wade Carpenter: No, you can't. And I guess that sort of brings up some thoughts I've had lately with us moving to a new office and throwing out some of the 10 keys. It was, had to be on our desk for first 25 years of my existence. I couldn't live without a 10 key, and now I don't even have a calculator on my desk. I mean, it's all like in Excel or all these programs. How do we get there? One of the things I wanted to talk about is sort of like getting these tools to work.

Actually wanted to tell a little story. About two weeks ago I got a prospect call. She booked a Zoom meeting and I get on there, I found a little bit, she was actually in Atlanta, relatively near me. Found out a little bit about them. Got on the call with this lady and she did not speak English, strictly Spanish. She's like, you speak, speak Spanish? You [00:04:00] know? And it's like, okay.

She wanted help with her construction accounting. She's like, well, you got a translator? No, she's in wherever she said she was. And I was like, okay. So we sat there 30 minutes. I was like typing stuff in ChatGPT, trying to figure out how to use Google Translator real quick. And it was a fruitless conversation.

She wanted my help and I really wanted to help her, but we were not speaking the same language. And after 30 minutes of doing this, it was like, send me an email. Let's get it together when we can have somebody translate.

I know that's sort of a metaphor here, because every day in construction finance, we have estimators, we have project managers, CPAs, you got your bond agents, maybe your bankers.

They may be all fluent in their dialect, but we're not speaking the same language. And part of this goes along with an article you sent me that we'll get into a little bit about streamlining AI, streamlining Work In Progress, and all speaking the same language, but I think right now it's sometimes a language barrier. Does that make sense? 

Stephen Brown: Yeah. It's absolutely a language [00:05:00] barrier I get it. Your real life example was factual. Two different languages trying to communicate and help each other. And then you're expecting AI to be the universal source for fixing all those problems. I also can't help thinking about how many of our listeners that aren't in the United States are having to translate our podcasts over to understand what we're trying to share with them.

Wade Carpenter: Yeah, and I'm sure Google can translate some of that kind of stuff. It's more about, how are we getting our systems to talk to each other? I guess all this stuff happens to me sometimes and I start thinking about things.

But my wife recently had knee surgery and she was at physical therapy. This was like her third round of physical therapy. So we go in there and they give 10 pages of things to fill out. And they're the exact same thing that she has to fill out multiple times, and it's all paper based. My wife was like, why do I have to fill this out? Especially like all the medicines and all this stuff. I've already done this. I said we just have to do it. I was just sitting there, I was like, ask the guy don't you have to take this piece of paper and type it in your system? [00:06:00] Yes. that make any sense? Could you not give us an iPad? We fill out a form.

Sometimes we're doing some things and they don't make sense. They're not-- just like the translation. We need to make them talk to each other. Digitized doesn't necessarily mean connected if we're talking two different languages. We have stopped using things like fax machines. We used to live off of fax machines, you remember? People stopped using copiers. We used to burn up a copier, and now we don't even have a copier in our office. We just scan everything.

I guess where I'm going with this is, how do we get things in a form that are easy to work with. There's a reason I'm going through all this. But how do we get all this manual data entry into data that the people that need it, in their hands as fast as possible, and in a way they can read it and understand it?

Stephen Brown: Well, that's a great question. How do we do that? And first of all, I have to applaud you for getting rid of your calculator. You call it a 10 key. That had to have been hard. First thing that popped into my mind was, oh Lord, help me, you know, if Richard took away my copy machine. We have them all over McDaniel Whitley. [00:07:00] He's always said that that's the cost of having me as a business partner is how much paper I destroy.

And then I also think back about your wife at the doctor. My first impression was, okay, there's nothing that can be done about that. Forget about it. 

Wade Carpenter: I don't know about that. I still feel there's privacy concerns and all that stuff. But my doctor has a thing. It's like an iPad. We fill stuff out and all this, and I'm not handwriting a bunch of stuff, and then you don't have somebody on the other end keying all that in.

But I just wanted to go back, so we've got a translation problem. We've got systems that like paper systems and digital systems, getting them to talk to each other. And I also got to thinking about how AI basically fixed its own translation problem. And I'm gonna geek out here, but if anybody that's listening to this knows anything about what a MCP server is it stands for model context protocol, and don't forget the acronyms, but it's a way of these AI systems to talk to another system. [00:08:00] Anthropic came out with it. And it allows you to talk to your calendar and your email and all kinds of different programs, where there are programs in the APIs-- and again, I'm not trying to geek out, but there are different protocols and you have to learn each protocol.

Well, MCP servers basically made it so that we can translate these things together. AI can read it in a certain way and interact with it. So in a way, AI had the same kind of problem. You think back to ChatGPT, we'd type something in and get some kind of result out and then we'd have to copy it and then paste it into our Word document, and it didn't really do anything to begin with.

There's a lot of things that it is starting to do, but that MCP server part was a pivotal thing where AI is like, Hey, we're gonna be able to translate Google Docs or Google Sheets to Airtable or things like that.

Stephen Brown: How have you made that work in your business? How have you used this protocol?

Wade Carpenter: There's a deep dive we could do on that too. But the point of all of it is AI really had to learn the same lesson we really need [00:09:00] to learn in construction. We can't have all these different systems, and everybody talking in different languages.

One of the genesis of this whole conversation was an article you sent me about again, I'm gonna geek out, but it's this XBRL. It stands for Extensible Business Reporting Language. It was about Work In Progress. XBRL has always been sort of like a translator and usually used for stock exchange type companies, really big companies.

In my mind it just is reporting like your balance sheet and your PNL in a way that you know is universal. But when I saw this article, I was like, okay, WIP. How does that really apply? I'm gonna ask you, Stephen, since you're in the bonding world, you've probably got a WIP template that you give your client to fill out sometimes, right? 

Stephen Brown: Yes.

Wade Carpenter: Does it have McDaniel Whitley Bonding and Insurance on it? 

Stephen Brown: No. I mean, ideally that doesn't do much good with the bonding companies. That WIP needs to tie in with a balance sheet.

Wade Carpenter: No. But my point is, you have your template, right?

Stephen Brown: Right. And it's a self [00:10:00] calculating blank. It helps figure out how to fill out a WIP Yes. 

Wade Carpenter: So if you go across town and find one of your competitors, do they have their own template?

Stephen Brown: I'm sure they do. And it is probably the same template that has been handed around by everyone. But nevertheless, they may have a completely different template. 

Wade Carpenter: Well they may. And they probably have their different ways they report it. And again, I thought this is silly, but somebody might be calling it on their template, okay, this is your under billings. One may say cost and estimated earnings in excess of billings on uncompleted contracts.

The terms that we use can be interpreted different ways. And what this protocol was supposed to be doing is like, how can we standardize WIP? Because, how do we define billings? What goes in that? Is retainage in that?

And one of the biggest thoughts that the article really didn't address, it is probably just one of my hangups is reporting a loss on a contract in progress.

I know this is a problem that most people really don't care about, but Generally Accepted Accounting Principles say [00:11:00] we're supposed to report a loss as soon as we know what it is.

I worked at a very large firm in Atlanta. One of the top firms in Atlanta that did not know how to report this correctly. It was going through Over Under Billings when it's actually supposed to be a component of cost. Even CPAs don't know how to translate this the same way.

My thought with this whole thing is okay, WIP is going you know, with this standardized reporting language. I've got my template and you've got your template, but we could get template that meant the same thing that you could send to all your underwriters from across different sureties, and it would mean the same thing.

Stephen Brown: And if that template could tie into our customer's accounting system and help them produce the report, that would be even more wonderful. Because, you know, it's not supposed to be our job to produce a WIP. You should have an accounting system that does that yourself, and you use it as a tool. So when we do a wip, as your bonding agency, we're gonna mess it up and there's gonna be a lot of questions that don't make sense that the bond underwriters have because, the data doesn't always [00:12:00] tie in with the balance sheet and income statement. It can't.

Wade Carpenter: Yeah,

I mean, imagine if our reports had a translator. Like where you're in QuickBooks or Procore, or your bonding portal, they had them all speaking the same language. That's exactly where I was going with this. We can talk about some of the nuances of that, but I think we'd lose everybody maybe in the translation.

 My thought experiment today, what I'm hoping if anybody's still listening to this, is what if we can make all these systems talk to each other? What if under billing meant the same thing to a CPA that it does to the bonding agent, the banker, to the contractor without having multiple email chains explaining what-- I mean I'm sure when you do write up to your underwriters, can you relate to that?

Stephen Brown: Absolutely. It would be a glorious thing. Because how the underwriters interpret the financial data that they get and how you interpret it are two different things. And as bond agents, you gotta be prepared defend your customer. If the underwriter is interpreting that data as, as something that's gonna hurt your ability to get their bond approved. That's [00:13:00] huge.

You were talking about that and I didn't tell you this, my son Jim was working with a company literally they used 2005, QuickBooks. So that's not QuickBooks online. It was brutal to extract the data and put it in a manageable understanding spreadsheets and other things. He introduced to his new firm this software that they had used at the bigger firm. it's extremely expensive. And he said, I went into my partner and then we went into the head honcho. We explained what it was we wanted to accomplish and what the cost was, which was tremendous. he wanted to know how many hours it was gonna save. And we told him and He said, yeah, yeah, we need to do this. Go ahead. And that was the end of the meeting. Isn't that amazing? He had to spend all that money to be able to understand some data and interpolate it from 2005. And to be able to warrant that, that date is accurate. It's scary. I don't know how you do it, Wade.

Wade Carpenter: It's a concept problem, but whether it's in accounting, whether [00:14:00] it's WIP, whether it's just more a problem like, okay, how do we do estimates or purchase orders or having everybody talk the same language. Your estimators talking to your project managers, talking to the owners.

The bigger problem here is in my opinion, construction is a little slower sometimes to adopt some of these technologies. I was amazed at this particular person that is getting on up in age and he was like, I'm trying to figure out how AI's gonna fix my world. Their systems don't talk to each other to an extreme, I would say. Just sort of like your 2005 situation.

I go back to that episode many years ago when we talked about stepping over the toolbox. If it takes you an extra 15 minutes to fix something that you do every week, if you constantly go around this hurdle.

We don't necessarily have to spend tons of money on some of this software to be able to fix some of that stuff in the age of AI. But how are we getting away from this translation, friction? Making things talk to each other so that we can eventually get to where all these things really, whether we all completely get to the point where AI actually is [00:15:00] running the show, I don't know. But what do you think?

Stephen Brown: I think it's fascinating. I think it's something we ought to continue to explore, and give some advice about as we can figure it out and as our listeners figure it out, let us know. Because we're all pulling with the same set of orders here, aren't we, Wade? 

Wade Carpenter: Yeah, and again I could probably geek out on this episode quite a while because there's a lot we could talk about on this subject. But at the end of the day, the real cheat code I guess I would say, is AI can't fix chaos. It can only basically amplify the clarity. But if you got same thing, garbage in, garbage out.

So ultimately what we need to be doing is thinking about how we can make our systems streamlined, number one, but number two, understand them. We've had a few episodes about FP&A and all this stuff, but we can't get there if we don't have the systems set up to be able to do that.

Stephen Brown: Yeah, that makes perfect sense.

Wade Carpenter: If nothing else, maybe I made some people think today, what is that constraint that's holding you back in your business?

You're going into 2026, maybe you've been hoping that next AI revolution's [00:16:00] gonna fix everything in your business. But sometimes we have these translation problems and I think the company that eliminates some of this translation friction, those are probably gonna be the ones first to win, I think.

If you have a standard that everybody follows, if there's a standard for the way you do estimates or whatever that everybody else does the same way that's just the building blocks for making all these AI systems talk to each other. 

Stephen Brown: It makes perfect sense.

Wade Carpenter: Okay. I guess this is probably a good place to close this one up. If we made you think about any of this stuff, I'd love some comments on this, because this is something I've thought about in my own business, but it's conversations I'm having with my customers and clients as well.

If you have any thoughts, put them in the comments below. We do this every single week. We appreciate it if you'd share, subscribe, do all that stuff, it always helps the channel out. And we will see you on the next show. 

[00:17:00]