Introduction
Ryan: Welcome, everyone, to the AiFounders podcast show. Our podcast is dedicated to celebrating the remarkable accomplishments of AI innovators, entrepreneurs, and visionary founders, captivating their stories behind the movements. I am your host, Ryan Davies. I have the honor of hosting today’s AI in construction and maintenance, focusing on improving training, quality, and production with our special guest, Matt Kleiman.
Matt: Hey, Ryan, great to be here. Always enjoyed speaking with you.
Background and Journey into AI in Construction
Ryan: This is such a great topic. Your expertise is unparalleled in this area. I can’t wait to dive in. For those just getting introduced to Matt, he is the co-founder and CEO of Cumulus Digital Systems, a company dedicated to digitizing quality insurance and productivity efficiencies in industrial maintenance and construction. Matt is a thought leader in industrial IoT, connected worker, connected workforce, digital transformation, and construction technology. Congratulations on your new book. We’ll discuss that further. Well, Matt, let’s dive right in. We’re talking about the construction and maintenance industry, which is incredibly technology-heavy. However, despite that, it remains a legacy industry, relying on pen and paper. Tell us about your journey and what inspired you to bring AI into these legacy areas successfully.
Matt: Thanks, Ryan. You’re correct that it’s a legacy industry crucial to the global economy. The technology challenges are unparalleled, and it’s vital for our future. However, it remains an old-school industry, using analog systems and lagging in digital transformation. My journey started in aerospace, and after the Deepwater Horizon disaster, Shell hired me to bring capabilities from aerospace to enhance safety. After six years, I left to start Cumulus. The industry faced challenges in adopting AI products due to the lack of accessible data. Companies asked for data, but less than 10% of construction was in usable form. This led to the creation of Cumulus, focusing on digitizing Work at the individual worker level to create structured data. With the release of new LLMs, we can now make sense of human language in construction documentation, enabling the creation of digital workflows.
AI in Construction – Improving Training, Quality, and Production
Ryan: That’s perfect. You touched on the challenge of different languages in the construction industry and the need to improve communication. The focus of AI in construction extends beyond productivity to areas like safety and training. Can you paint a picture of how AI can improve the grand scale in various aspects?
Matt: AI is valuable wherever a person needs to follow a process. In construction, from building houses to complex structures, inspections and processes are critical. This extends to every business, like a bakery, ensuring daily production. AI helps streamline these processes by making it easier for individuals to gather and interpret information, improving efficiency in the last 10 percent of their tasks.
Ryan: Perfect. Yeah. And it eliminates that interpretation gap that might be there. You have different people who see things in different ways. Looking at safety standards or whatever it is it eliminates the gray area and really drives it down to that point. As you mentioned, it gives the ability for a company to be more productive, putting workers to work more in safer environments and scaling up their business in ways they probably didn’t know were possible. So, with that said, this data is all-important. You mentioned this is really legacy stuff. We talked before the podcast about how you’ve got people with Sharpies on butcher paper whiteboards behind them that fade into existence over time. How have you been able to collect and arrange the data to help the industry in such a profound way?
Matt: Well, the first thing we do is start with making the focus on the procedure the worker has to follow. What so often happened before was that people would collect random checklists, photographs, or drawings. They make sense to somebody but seem random to a neutral observer. Or I’m going to do this checklist, and they needed to be organized around a central controlling idea or theme that an AI or even a human can then make sense of. We’ve really made the focus of the job site being the worker, capturing what the worker is doing in a structured way and then building out from there. Once what the workers are doing, you can understand your schedule, risks, inspections, and project completion. There’s a saying in construction that the last 10% of a project takes 50% of the time because there’s Work to make everything come together at the end. Suppose you have true visibility and structure around what the worker is doing and tie everything else to that. In that case, you’ve taken out a huge percentage of the variability on projects that lead to delays, over-budget, rework, and accidents. It’s finding the central governing theme, creating structure, and building up from there.
Data Collection Challenges and Solutions
Ryan: So with Cumulus specifically, or whether using something like Cumulus or someone’s trying to put this in place for themselves, give us an example of what that walkthrough looks like. Tell us what that process ends up looking like on the other end. This is an ability to digitize processes in seconds and minutes instead of groups of people doing it over weeks and months. So tell us what that process looks like on the other end.
Matt: One of the first projects I worked on in the industry highlighted the organized or disorganized chaos. There was effort in planning the Work digitally, but when it came to the field, everything went back to paper. Workers had detailed digital work plans, but they used paper to capture their activities. Then, there were clerks transposing the paper records back into digital systems in trailers weeks later. This led to mistakes, delays, and issues with the accuracy of the information. Cumulus focuses on making the worker the central point, capturing their actions in a structured way, providing true visibility and structure. This digital-to-digital transition eliminates the loss of information that occurred in the digital-to-analog-to-digital process. It addresses the gap in systems thinking, making projects more efficient. In the early days of Cumulus, when some of our customers first digitized their workflows, they would take their big engineering documents. These are large companies, not small contractors. They would take their carefully crafted procedure documents and turn them into digital workflows. Once they see it, they realize these make no sense at all because the engineers were thinking like engineers, not as people who have to execute something in the field. In fact, it makes a better result because there were just things that have been built in over time that just made no sense, and you only appreciate that once you have it in a digital form that someone can easily visualize.
Ryan: It’s one thing to walk through it and see it and almost robotically go through it. But when you can take a step back and look at it, you go, there’s redundancy, and that doesn’t make sense. I don’t even know why we do that anymore. This gives you a place to go. It’s really hard to grow if you don’t have a foundation to work from. This gives you that foundation. Let’s talk about the book Work Done. So, I know it just came out. It’s available on Amazon and other great retailers. One of the things you talk about in there and one of the things I always hear on AI in this podcast, and over and over, regardless of the industry, is, how do I get started? It’s so expensive. I’m scared about the upkeep if I don’t know what I’m doing. Versus what’s the cost of me not doing anything, the cost of inaction, and staying a laggard? Let’s talk about that. The construction industry specifically loves traditional methodologies. We’ve always done it this way. People are swinging hammers. We got people that are very; this is how I’ve always done it. Let’s stick to it. Talk to our audience about that. How do you start to calculate that cost of what if I don’t do anything that I can’t afford to? And what you’ve seen.
Matt: Understanding the true costs of the status quo is incredibly important, and most companies are really bad at it because what has happened over time is that some things are just considered the cost of doing business. It’s just the way things are done. We don’t even account for them anymore. They’re these hidden costs that nobody is responsible for because they were hidden. They were not directly able to account for them. But if you think about delays in a construction project, what are all the costs of those delays? Not only do you have to pay the workers longer, you have to keep your equipment rented for a longer period. But then also, you’re delayed in starting up your structure, whatever you’re building. Let’s say you’re building a manufacturing facility. Well, there’s the cost of not being able to produce whatever goods that manufacturing facility is producing. There’s the cost of misinformation. So again, going back to the digital analysis, there’s the long time people spend finding information on your job site, walking back and forth from the trailer to the field because they’re just trying to figure out what they’re supposed to do or resolve a conflict in information. What is the cost of that? And how does it all add up? Most companies need to improve at accounting for it because it’s just bucketed into, well, that’s the way things are done. So, in the book, we talk about different ways you can think about those costs and quantify those costs so that when you’re looking at the ROI of doing something new, whether it’s AI or drone surveys or whatever it might be, you can have a true understanding of the benefits of that new technology, or you could also understand. Well, new technology isn’t always the answer, and this is talking against my financial interest as a technology entrepreneur. But a lot of times, it’s not a technology problem that these companies have. A lot of companies need to improve at procurement, for example. They need to improve at managing their supply chain, and digitizing poor procurement practices isn’t going to fix those procurement practices. And to give an example, if you’re paying all of your labor providers an hourly rate and they make more money, the more they work. Well, no wonder things are going to take longer than they should. No wonder there’s going to be rework on the job site. No wonder they’re not going to be efficient if you don’t put in your contracts incentives for them to be more, where they could actually make more money if they deliver on time and ahead of schedule, which almost never happens in these contracts. Well, it’s amazing how human behavior will change once the incentives change. That’s just an example.
The Future of AI in Construction – Short and Long-Term Perspectives
Ryan: That’s amazing. I always love this question. So, it’s crystal ball time. Where do you see the industry evolving in the next short and longer-term periods? We see these tools being picked up, and they’re starting to see, I love your comment, and you’ve made it before with me, that AI isn’t going to replace people; it’s going to make people more efficient and better. The people who embrace it faster and the companies that embrace it faster are going to get ahead and stay ahead and scale up, right? So, with that in mind, you see these companies that are embracing it. What kind of results do you see now and into the future for them in terms of that?
Matt: The biggest impact will be, I mentioned in the study before that said, less than 10% of data that’s already generated in construction is in a usable form. Well, what if you can, with LLMs and other technology, make that other 90% accessible and understandable, where it’s almost like you’re standing in a dark room, you have a small flashlight, you see just what’s around you, and then you turn the lights on, and you see everything. That is going to be incredibly powerful. With this data, you’re going to be able to see, learn, and understand things. The companies that are forward-looking and embracing these new technologies with that in mind, not just whatever the shiny technology is. Oh, we want a chatbot. Great. Do you want a chatbot? But what is the chatbot using as its data source? If it’s just using contract language, that’s fine; that could be valuable, but it has to be everything on the job site. The LLMs can then understand it and help project managers make much better decisions and much better predictions that will tremendously increase the productivity, safety, and effectiveness of workers on the job site.
Ryan: With all of that, I want to, we’re getting towards the end here, but I want to give a chance for you to tell people about where they could pick up Work Done, right? And more about that and Cumulus Digital Systems. Most importantly, people can tap you on the shoulder to get more information and learn more about you and how they can bring some of this to what they’re doing, whether it’s in the construction industry or otherwise, seeing the parallels.
Matt: Thanks, Ryan. As you mentioned earlier, it’s available on Amazon. Just search Work Done with my name, Matthew Kleiman, and you’ll find it. I am very active on LinkedIn. That’s the social network for easiest to find me to message me. Find me there. Our Cumulus Digital Systems is Cumulus DS as in David Sierra cumulusds.com. Check out our website, and our new AI app is actually dropping into the app stores; that’ll be on November 9th, 2023. Download it, play around with it, and let us know what you think. We are really excited about the possibility of this technology. We are really looking for feedback from people using it, both good and bad because it’s like anything new; there are going to be pluses and minuses, and it could only get better.
Ryan: You really are a master of productivity. Book, October 31st; app comes out November 9th. I don’t even want to know what’s on the horizon for you. You probably have something going on.
Matt: I have an awesome team, and the credit really goes to them for the hard work and the expertise that they bring to and the passion for solving these problems.
Ryan: Thank you so much, Matthew. I know our listeners got a lot out of this. I always get a ton out of our conversations together, and I hope we can have you back again. We’ve got so many other topics that we can talk about in terms of where we want to go and what we could do in the AI space. We just really scratched the surface today on what you do and what can be done in the construction and other industries.
Matt: Thanks, Ryan. It was a lot of fun, as usual, and I look forward to next time.
Ryan: I can’t wait already. I want to thank everybody for joining us on this enlightening journey through AI innovation, and I hope that you’ve been inspired by the incredible stories that were shared today. And remember, the future is driven by pioneers like our guest, Matthew Kleiman, and the limitless possibilities of AI. So stay curious, stay innovative, and keep exploring the boundless horizons of technology. But before we sign off, we have a small request for our dedicated listeners. If you enjoy the podcast, please take a moment, share it, tell people about it, leave a review, and subscribe to the show on your favorite platform. It’s your feedback and support that helps us bring you more amazing content and more amazing guests like Matt here today. Until next time, this is Ryan Davies signing off. Thanks, everybody. Take care out there.