AI Powered Construction Estimating

Podcast Transcript

Steven Rossi: 

Hello there and welcome to the Construction Revolution Podcast. My name is Steven Rossi and here on the show we explore the latest trends, technologies, people, and organizations that are revolutionizing or disrupting the construction industry and are changing what the industry will look like tomorrow. 

Today on the show I’m speaking with former Florida Congressman Patrick Murphy, the founder and CEO of Togal.AI. With expertise spanning from working hands on in construction, to serving in the US Congress, and now leading a pioneering AI construction tech firm, Patrick brings a unique blend of expertise and experiences. Join us as we delve into Patrick’s extensive background and his vision for integrating AI in the construction industry. Tune into discover how Togal.AI is revolutionizing estimating by automating the drawing analysis process, allowing for estimators to spend their time on more valuable tasks. 

Hello Patrick, welcome to the Construction Revolution Podcast. How are you doing today? 

Patrick Murphy: 

I’m great. Thanks for having me today. 

Steven Rossi: 

Yeah, absolutely. I’m excited to learn more about Togal.AI and about your journey to starting it. 

Patrick Murphy: 

Likewise, I’m looking forward to the conversation. 

Steven Rossi: 

I guess let’s get started. If you can just give us an overview of what Togal.AI is so that we can get into it deeper. 

Patrick Murphy: 

Sure. Steven, if it’s okay, I might just do a quick background because it kind of segues right into Togal, maybe helps explain it a little bit better, if that’s all right with you. 

Steven Rossi: 

Absolutely. Sounds good. 

Patrick Murphy: 

Yeah. I was raised in a family construction business. We do a lot of vertical construction in Florida and worked in all aspects of the business. I was a day laborer, project manager, worked with my grandfather doing estimates back in the day with the scale ruler and rolled up plans and worked in all aspects of the business. Then college, got my CPA license, worked at Deloitte for a bit and then ended up serving in Congress for a couple terms in Washington DC. 

I rejoined the family business after being gone for about 12 years and we’ve been blessed to be in a great market in Florida, a lot of construction, the company has grown. I realized when coming back that basically nothing had changed the same old processes and same old inefficiencies. As I’m sure you know and you’ve seen, there’s been very sort of little innovation and there’s been numerous studies from McKinsey and others showing that productivity has basically remained flat in construction over the last 50 years. 

I started really taking a look at how we could improve our own efficiency in our construction company and went to our financials and saw that the single biggest piece of overhead in our company was pre-construction estimating. The single biggest piece within that was takeoffs, right? Doing the square foot analysis, the quantity analysis. I was reminded that about 50 to 60% of our team’s time, 30 plus estimators, are doing quantity surveys, a great deal of their time, basically coloring for a living. I said, “What are we doing in today’s world? There’s got to be a better way.” 

You asked a simple question, but this is the background. The thesis of Togal was to automate the takeoff process. How do we compress days or weeks of takeoffs into seconds using artificial intelligence? If a human can look at a set of plans and determine the difference in a bedroom and a bathroom and a hallway and the difference in a wall and an arrow say, then so can a computer if it’s trained correctly. 

We spent about three years aggregating literally hundreds of thousands of unique floor plans from around the world of all different languages, labeling them in a very, very unique precise manner and then developing the layer of proprietary algorithms on top of that to do that automation and come up with exactly the corner where you should be doing, just say if you’re doing a flooring takeoff versus a wall takeoff versus account versus an electrical takeoff, et cetera. Anyway, we have really rethought the entire takeoff process. We’re a web-based takeoff tool that has reformed the way you upload your plans, the way you label them, the way you manage them, the way you can share with collaboration, using AI for the takeoff, and then exporting in all sorts of different ways. Really rethought the whole process and I’m biased, but believe we brought a lot of efficiencies to the process. 

Steven Rossi: 

Yeah, absolutely. Yeah, that’s a great story. I mean, you’re absolutely spot on about how innovation is often lacking in construction and whereas even when companies like yours and ours come along, they can be slow to be adopted as well, which I think is another whole separate issue. You touched on, I guess this started as sort of a pain for your own company. What kind of companies are your target market? Can you share a standout application from one of them, even if that was your own company? 

Patrick Murphy: 

Yeah, so when we started Togal, originally we thought it’d be really just for Coastal, that we could save a few million dollars a year not doing takeoffs and expedite the process. As we started building it, seeing it work, we realized this could apply to companies of all sizes. We spun the company out, completely independent standalone company, and then we went and started talking to other contractors of all sizes to get their input to make sure we were building a company that could serve the biggest contractors or the smallest in the world. We have companies, one of our biggest right now is Clark Construction. They’re, I don’t know, 8 billion a year plus. We’ve got 140 licenses with them right now. That’s one of our bigger clients. We’ve also got mom and pops doing three to 5 million a year with one or two employees using Togal with equal success and outcomes. We’ve tried to ensure that the platform is intuitive and streamlined and easy to get up and running. There’s no long onboarding. It’s 20, 30 minutes to learn the tech, very intuitive. 

We have spent time of course focused on the big enterprise customers, but I can make a pretty good argument that those mid-size and smaller contractors could see almost more value in Togal because as you’re trying to grow as a construction company, you got to bid on more work. In order to bid on more work, often you have to go hire an estimator, $100,000 plus benefits and everything else. Instead of doing that, what we believe you can do is get Togal for $3,000 a year and bid on more work, get those bids out the door, and now you can have better bids because your team is going to be focused on the higher value tasks, the value engineering, and the scoping and the bid leveling and all the things that really help you win more work as a contractor, instead of coloring for 50% of the time. 

Again, scalable to companies of all sizes. Coastal Construction has, I think about 20 licenses with Togal, and they saw about 10,000 hours of savings, which equated to about a million dollars in their first year using Togal. Definitely quantifiable, basically instant results in using the platform that again, is scalable to general contractors to trades of all different sectors. 

Steven Rossi: 

Oh, that’s awesome. I know estimating is varying in a lot of pre-construction, but estimating in particular is very labor-intensive, so it’s great that you’re able to save time and not have estimators color, as you put it, which a good way to put it. Apart from the coloring part, I’m wondering if you can go into a bit more detail on how Togal.AI differs from traditional estimating software and estimating processes. 

Patrick Murphy: 

Sure. So I’ll start really just from the beginning, which is uploading the plans. Being web-based, the uploading of the plans is seamless. We have a great filing system and automatically label and name all of your plans so you don’t have to spend the time resorting that. Oftentimes estimators will go to different applications to do that in and then bring it back so they have a clear legend. We do that automatically. You’ve got the sorting, the uploading, the labeling. 

We have a collaboration tool, which is really interesting. If you have folks maybe working from home, working remote, if you have multiple offices or just want to make sure that everyone’s working off the same set of documents, our tool is great because it’s like Google Sheets, but for takeoffs. Endless amount of people can have different profiles and approvals for what they can do, different permissions. Maybe just say the boss can do everything in everyone’s work, but maybe a more junior person can only count doors that day and that’s all they do. Then someone does floors and someone does walls and all that can be done at the same exact time and then the boss can see all that work happening. 

We also have a comment feature right on there. Whether it’s an internal team member or maybe a subcontractor or a third party you’re working with. If you have a question about something, you can just put a comment right there and they can see it live so you’re not emailing, “Hey,” so-and-so, “Here’s the attachment, go to page A.232 and seeing the bottom right corner, this feature,” and then they’d have the wrong version and then you got to find the right version. There’s all this wasted time, it’s all live, it’s right there, you can comment, you can have a whole conversation and get problems solved quicker. Another great, I think attribute of ours. 

Of course, the AI capabilities. We process any file type, PDF, CAD, JPEG, anything you upload, we’ll process. The user just has to set the scale and click what we call the magic green button. You’ll get your takeoffs, your net area, your gross area, your linear takeoffs, and your counts. We only count a certain amount of items automatically. Then we allow the user to go and draw a bounding box around say an electrical outlet or fixture or symbol or tag, maybe a word they’re looking for. We will find all instances of that with 99% accuracy. Then you can make small modifications to that to continue finding more symbols maybe that are similar but a little bit different. It’s easier to show it than explain it, but makes all that process a lot easier. 

Then the exporting, a lot of the sorting and quantification and classification can be done right in Togal. When you export it either to your own proprietary estimating software, your Excel sheet, maybe it’s a third party estimating company you work with, we can tie it directly to that. It’s always consistent and you can have the same takeoff show consistently throughout your jobs. 

A lot of exciting features. One more I’ll throw in there is TogalGPT. I guess it’s maybe self-explanatory for some. We partnered with GPT and built a custom version that is suited for construction documents. Number one, users upload all of their documents from their plans to their specs to their RFIs, to the schedules and budgets and everything they have, thousands, tens of thousands of pages get uploaded to TogalGPT. Then you just have a conversation with it. If you want to know who’s responsible for installing the windows on the 10th floor, you can just ask who’s responsible for installing the windows and who did the caulking on the interior, what’s the appliance package and the penthouse balcony, whatever the question is. Instead of having to go and read the contract and the scope and send it to the attorney or all that back and forth, just have a question. 

Two important things. Number one, we have basically eliminated all of the hallucinations, all of the sort of made up information that you would get on GPT normal, where they’re looking and scouring the whole internet. We only search what was uploaded by the user and if the information isn’t there, we tell you that. We’re not dreaming something up. Number two, it is completely secure and cut off from OpenAI and Microsoft. They can’t use your data to train and improve their algorithms. They’re doing that on their own with plenty of other data, they don’t need ours. We cut that off so it’s very secure. Exciting, cool feature. We have some clients using it for scoping, for bidding. We have some using it in the field. We’re releasing a mobile friendly version, basically an app. Users can be walking a job site and just ask those questions as they’re maybe talking to a subcontractor, talking to a client, whomever it might be to get those questions answered on the spot, whether it’s scheduling or budget or something else that they need to know. Yeah, a lot of exciting features in the platform. 

Steven Rossi: 

Yeah, absolutely. It sounds like from beginning to end, you guys thought of everything. It’s very easy process and the GPT integration, it sounds really cool. I’m sure a lot of users must get a lot of benefit out of that. I’m wondering though, you mentioned the security issues with or concerns that people may have with GPT, but I’m wondering when you go out and pitch this to people, your software is very reliant on machine learning, AI, and ChatGPT, which especially in the construction industry, people aren’t always quick to jump on that. What sort of feedback, and do you ever get any pushback when you go out and pitch that to people? How do you overcome that? 

Patrick Murphy: 

Great question, and it kind of goes back to the beginning of this segment about the lack of innovation in the industry. I think there’s a lot of reasons for that. I think a lot of the innovation over the last couple of decades really hasn’t been that good or tailored to construction. A lot of folks in Silicon Valley, whatever, they don’t understand the nuance of construction. When you talk to them about it, their first reaction is, “Oh, it’s so simple. These buildings kind of build themselves. You stack some blocks and add some drywall and it’s pretty easy.” I’ve had these conversations with people with, “Oh, it’s not complicated. The AI on plans is easy.” Well wait until they try it, right? Wait until they realize the amount of nuance in a set of plans no matter what you’re building and the diversity of those plans. You all know this, but there’s some real good reasons for that. 

Number one, most of what’s built is unique, right? We’re not just building the same old Boeing 737 over and over and over again and changing it every 10 years. It’s completely different. Every time you do that, there’s a new engineer, there’s a new architect, a new interior designer, maybe a new builder and subs and maybe a new building technology that’s being implemented. It’s always different. On top of that I should say, is there’s not a lot of standardization in say electrical outlet symbols or fixtures. It all starts with architects that are artists at the end of the day and they’re designing a dream for the client. Because of that, there’s a lot of nuance. I know you and the listeners all know this, but just reiterating that I think is important because the technology hasn’t been there to process the amount of unique information. 

However, because of AI, because of some of the recent breakthroughs, it is finally affordable in my opinion, to use AI. We’re starting to see the models that can handle the millions or endless amounts of scenarios, process that instantly and give back the user some actionable information. That’s the big difference in why I think AI is transformative for our industry because it’s really good at taking all that unique information and now giving the user some output. I think we’re going to see a lot of changes and more adoption. What we’re seeing, to answer your question, is yes, there are sort of the salty crusty folks that don’t want to see change. Maybe they’re toward the end of their career often and they’ve been doing it the same way and they just not ready for that. That’s fine. There are a lot of folks that we see that do want to get more efficiencies, that do want to grow their margins, that do want to make more money and a lot of younger folks joining a construction company. 

I know we see it, estimators, engineers, they don’t want to color for 50% of the day. They want to create value, they want to value engineer, they want to use their brains for the creative thing. Let the computer do the mundane boring stuff, whether that’s searching documents with GPT or whether that’s determining that the gross square footage of a 10-story building, let the computer do that stuff. Again, we’re not going to get everyone to jump up and say, “Yes, let me try it,” but more often than not, people are willing and open to give it a shot. 

Steven Rossi: 

Yeah, sure. I think that’s a great point that having technologies like this that are tailored and built specifically for the construction industry by someone in the construction industry, it changes everything as opposed to having things sort of being like, “Oh, this could be a use case. Let’s pitch this to construction,” and then it doesn’t work very well. You mentioned earlier, you mentioned it took over three years and I’m sure countless drawings and things that you analyzed. I’m wondering what that process was like. You mentioned you sourced, obviously you must have had a lot of drawings on file from Coastal, but how you sort of went about getting all of that information and teaching the algorithm. 

Patrick Murphy: 

That turned out to be one of the real challenges and I think part why others haven’t done this, is just aggregating that data. The old saying, garbage in, garbage out. We quite frankly tried without knowing, tried to shortcut the labeling process to get to the end result quicker. We realized that there was no database that existed online or we could just go buy some label plans and use them and maybe there’d be a thousand plans that were labeled right here, but everything was different and it wasn’t consistent. We spent years literally, aggregating over 250,000 unique plans, again from all over the world, all languages, because we didn’t want to just say, “Get a bunch of apartments in the United States,” because it would be too consistent and might not do well on hospitality or schools or whatever else is out there. We want a really diverse set of plans, all languages and symbols because that would make us more powerful at the end of the day. 

We aggregated the data and then we hired 25 architects and engineers to do the labeling. Have a very stringent methodology to how we do that labeling of those plans and a review of those plans to ensure that they’re labeled in a very consistent manner. As we run our algorithms on top of that dataset, it’s constantly learning. Now as we have more users and more use cases, we can sort through that data, give more diverse plans with more results, and algorithms keep getting smarter and smarter and they keep learning. That’s the whole artificial intelligence component of this are the neural networks, nine of them that are constantly being run on our dataset to improve the results. 

Getting to 80% accuracy was a big deal, 90%, 95, 97, 98%, which is where we’re at now, getting those last couple points is very tricky. You need sort of order of magnitude more plans. We have to get close probably to a million plans and more accurate data to see those last points. I believe we’ve got a big headstart, really no competitors in this space. The more people use it, the more diverse plans we see, the more accurate we’re going to get. Getting to this point was years of work. 

Steven Rossi: 

Yeah, yeah, absolutely. That’s definitely with any program like this, the getting it off the ground once you are starting to collect input from users, that can definitely help accelerate the process. I think it’s a great point that it’s especially in construction, getting it that high nineties in accuracy because what we do needs to be so accurate that 80%, it’s not good enough. It’s crucial that it’s 99, in the high nineties percent. 

Patrick Murphy: 

Interestingly, what we did is we looked at humans takeoffs and discovered that humans in all sorts of different companies were about 97% accurate in their takeoffs. When we realized that, we said, “All right, 97% is our benchmark. We’re not going to market until we have 97% because that’s what humans are.” They might try to poke holes, but generally that’s where they’re at. That was sort of our threshold just to add color to what you just said. 

Steven Rossi: 

Yeah, that’s really interesting. I guess regression analysis and things like that, I’m wondering how your background with your CPA and working in accounting and finance and then also in politics helped you as a founder of this. 

Patrick Murphy: 

Well, look, we’ve got a long way to go and a lot of growth still. Obviously growing up in a construction business, seeing some of the inefficiencies, seeing some of the opportunities, I guess I would consider them, seeing how hard the industry works, knowing the hours that are put in, was sort of foundational for future things that I went on to do. Being a CPA, not the sexiest job, but definitely put in some long hours, learn about the importance of detail and auditing. I worked at a very large company, Deloitte, so got to see some of the big business processes. 

I ran for office for Congress because I was really tired of complaining. I was looking at a lot of issues from the debt to unemployment to healthcare to immigration, climate change, you name it, complaining with my friends. One of my friends sort of said, “Who the heck are you to complain? Do something about it.” I was like, “What am I, going to run for commissioners?” “No run for congress.” One thing led to another and I kind of said, “What do I have to lose? Why not? What’s the worst thing that happens? You lose, okay, so what?” Was lucky enough to put together a great team and won that election in 2012. 

Being in DC, I was the youngest member of Congress and was lucky enough to meet a lot of founders of some large tech companies and a lot of really large businesses and was getting exposed to things like artificial intelligence in 2012, 2013, before it was in the press a lot. Not that it wasn’t there, it wasn’t new, it’s been around since what the sixties, but it wasn’t common knowledge. I got to hear all these new great things happening in the world in innovation and technology, thinking about it really from a legislative standpoint or how are we going to prepare for the jobs of tomorrow from a legislative standpoint, knowing with the sharing economy and AI and all these big changes happening, we better get ready because our competitive countries are going to certainly try to beat us. 

I was thinking about it from that standpoint and then rejoined the family business, as I mentioned, and saw that not a lot had changed and there was some real opportunities to make these improvements. That background has helped me come up with the idea, but we’ve got a long way to go. First time CEO of a tech company, there’s a lot to learn and things move a lot quicker. There’s a lot more data that needs to be analyzed, should be analyzed, I need to get better at to keep yourself honest, keep the team honest. 

You might have a gut opinion about something and it might be completely wrong. The data could show something completely different. In politics, politicians, “oh, I need to say this on the campaign, this is what matters.” It could turn out it was anecdotal because your mom and wife or whatever said it, it might not be true. You go do a poll and you could see your constituents think completely different about an issue. Same thing in technology. You might have an opinion because that’s the way your company does it, but maybe not the industry and the data will show that, similar to a poll. 

I would also say the biggest similarity I point out in politics to startups is number one in politics, if you’re running for office, you have an idea, you have a message, you have something you believe in and you’re trying to convince others to believe in it, and you need to get them to work with you to volunteer to put blood, sweat, and tears in, you need to raise money. Then you have a primary, you have an election, and you see what people think of it ultimately. 

It’s kind of like a startup where you have an idea, you got to get people to believe in it. You don’t have a whole lot of money. You need to raise money. You got to get them to believe in that message and write a check, and then you go to market. That’s kind of like an election to me, where you do all this work and then you go to market and do people like it or not, and proof’s in the pudding. You either get people to sign up and pay or you don’t. Now there’s indications along the way where you’re beta testing, similar to maybe a poll that you would do in politics. You’re not completely flying in the dark. Then you have that election, you win, you get a product to market, you get some customers, but you got to keep iterating. 

You have to continue to improve your message in politics and get ready for the next election as you do in a startup. You can’t just be a one trick pony. You got to continue improving that product and get more customers and more people to like you and raise more money. You get told no a lot, just like in politics. You get beat up, right? Most investors and most donors tell you no. Sometimes you get them to believe in it. 

Steven Rossi: 

Yeah, absolutely. I would never guess. That’s an analogy and comparison, it’s great and it works really well and surprisingly well, I would say so. Yeah, that’s great. From going from politics, what made you want to get back into construction? 

Patrick Murphy: 

Yeah, I loved serving in Congress. I was humbled to be there and had a great team, great constituents, but I was really fed up honestly with the amount of dysfunction in our political system. I went in very naive thinking was policy and issues and debate, and it turns out it’s really a lot more about just getting reelected. I made a very conscious decision that I was going to go up the ladder, so to speak and run for Senate or out, that I wasn’t going to keep doing it. That’s no offense to the institution I respected, but it wasn’t fit with my personality well because I just don’t love the politics of it. 

Anyway, made that decision, wasn’t successful in that senate bid and was then thinking, “Okay, well do I want to stay in politics? Do I want to run for another elected office or get back to the private sector?” Ultimately that’s what I decided, and I’m blessed to have a family business, very lucky. I’m very close with my family. It’s one of the things I missed while being in DC, was that day-to-day interaction with my family. Having that opportunity was pretty cool. 

My interest was primarily in diversifying our business and we have a couple different ventures and investments that I try to lead up, this Togal.AI being first and foremost, but a few different exciting things we’re working on that I think will change the industry, that I believe could still be very impactful. If we can get more people into better quality homes for a better price faster, I’m doing my small part to make the world a better place. It might not be the big stage of Washington DC and Congress, but still feel good about what I’m doing. I think there’s a lot of cool opportunities in construction for us to help the world, to help our communities, and I think technology will really help transform and speed that process up. 

Steven Rossi: 

Absolutely. Yeah, I can imagine things move a lot faster a tech startup than in the legislature. 

Patrick Murphy: 

Yes, they do. 

Steven Rossi: 

Yeah. I’m curious, you mentioned you’re looking at some other things in the construction industry, I’m wondering, apart from estimating, what other applications have you seen AI beginning to be used? Are you excited to see it used in the future? 

Patrick Murphy: 

Yes, I think AI is going to really transform construction for the better and change the roles that we’re all used to having, I think in the industry. Like all industries, it’s going to change over time with technology. AI in particular will change construction more than most industries because we’re starting at such a low basis in my opinion, there’s so much room to grow, it could potentially be more impactful. 

I mean maybe just a few examples. We’ve talked about estimating on that front. Scheduling I think will definitely be impacted where countless amounts of scenarios can be contemplated with a live stream of data. As we are starting to connect more dots from shipping and delivery, actual job progress with the cameras we’re seeing and different sensors we’re seeing on a job that’s going to become a lot more efficient and give a lot more better information and predict various scheduling outcomes. I think we’ll see it there. On the actual maintenance delivery of jobs as well. I think we’ll have a lot more information and we’ll be able to predict certain outcomes based on historical data. 

There’s platforms that are doing some really exciting stuff in what they call sort of the no code AI, where companies are looking at all data points for a construction company, from the legal to the accounting to the plans, to the bidding and scheduling and project management softwares, everything, and coming up and finding inefficiencies in the business and seeing where a lot of the duplication is. Now this might sound like a dream to many, and it’s probably years before it really takes root, but there’s some interesting discoveries happening and some of the examples I’ve seen already of some really low hanging obvious inefficiencies in construction companies. 

I think ultimately generative AI will be the most transformational where, I believe in the next five to 10 years, there’ll be three inputs if you want to build a building, what’s the address, what’s the budget, and what’s the style? The generative AI will be drawing those plans and not just the pretty architectural drawing or elevation, but they’ll be drawing the MEP structural plans. There’s, I guess a few key reasons for that. 

Number one, there’s a lot of data out there and as I mentioned, we’ve aggregated hundreds of thousands of plans. Others are doing that. Again, not just for the architecture of the building but for electrical plans. If you could just say, “Get a million electrical plans,” that covers a lot of ground and most things that are built while might be unique in look and feel, it’s not necessarily the first time an electrical socket has been installed. There’s a reason, as we all know, why there’s a certain amount of electrical sockets. We know there’s a reason why there’s a certain amount of AC tonnage in a room. There are rules and metrics that are math-based usually to determine all of these things. The amount of structural steel, the spans of buildings, the size of a header, whatever, everything that goes into it, there’s a reason for it basically. 

AI is good at compressing those two things. Number one, the historical documents and historical examples, but also those rules and regulations and blending those two things together. Once you put your address in the computer can automatically determine what’s the code that they need to comply to, what’s the setback, what’s the density, all the look and feel that’s needed for that particular building. You put a budget in, it can help reign in the finishes and you put a style. Now it can go look at Mediterranean drawings or British colonial or contemporary, whatever it might be, and give you that look and feel and the rest can really be automated. 

I think we’re going to see complete transformation when that happens because then you start to see consolidation from your developer to your architecture engineers, GC subs, everything starts to blend into one. Look, I’m sure it’s going to take longer than I think, and it’s not just going to be overnight. This is a massive industry, trillions of dollars a year spent and tens of millions of people working in it. It’s going to take a long time to move that ship, but we’re on the precipice of that technology starting to work and we’re starting to see it. I’m really excited to see that feature in construction and hope our AI company is a small piece of it. 

Steven Rossi: 

Yeah, absolutely. I guess in the next five to 10 years, you mentioned generative AI and all this, is that sort of where you envision Togal.AI heading to, generating the drawings and then also analyzing them or where do you see the company headed? 

Patrick Murphy: 

Yep, great question. Right now our focus is architectural floor plans, or I should say it has been architectural floor plans. The last several months and next couple years, the focus is really going to be MEP structural, automating all sort the takeoffs and analysis of that full set of plans. Ultimately the user uploads that entire set and we can connect the dots between the electrical, mechanical architectural plans, for example, give all of those takeoffs and tie it all together. That’s sort of the holy grail in our opinion for the takeoff. 

As soon as that’s done, and simultaneously we are overlapping already, with building generative AI models for MEPs. I believe again that there’s going to be this transformation that takes some time. With our data set, I think we’re in a really good position to build those generative models that are, again, transformational for our industry. We’re trying to walk and chew gum and do a little bit of both while we are building these takeoff models, cognizant of the generative AI opportunities that are out there. 

What we found is doing an automated electrical takeoff, for example, it’s one thing to find the outlets and fixtures and all that, but determining all the runs that are necessary in those distances, is pretty complicated for the computer to determine one line from another. There’s so much, again, nuance and uniqueness in each electrical engineer’s plans, that skipping ahead to generative AI and predicting ahead of time what should be drawn and in accordance with the local code, I think could actually be faster and easier than trying to do a takeoff of what was already drawn. We’re kind of skipping to that in some of the MEP trades, especially mechanical, excuse me, and electrical. 

I’ve partnered with some folks that are in this space, that are in the trades helping to design some of these generative models. We’re focusing on single family homes first, thesis being that it’s a little easier to start there. If you can crack the code there, then we can expand it to bigger commercial projects. That’s where we’re going. In five to 10 years, I think it’ll be a full generative AI company. If you want to do a takeoff, fine, but we’ll be able to give the model right off the bat. 

Steven Rossi: 

Yeah, no, that’s really exciting and I can’t wait to see where you guys head, but I think it’s an interesting point of the approach to building something like this is, if the input is not something you can analyze well, why don’t we just generate the input, right? 

Patrick Murphy: 

Exactly. 

Steven Rossi: 

That’s a really interesting way to look at things. To wrap up, if people are interested in getting started with Togal.AI, where should they go and what does that process look like? 

Patrick Murphy: 

Yeah, thank you. Please go to our website, www.togal.ai. Togal is T-O-G-A-L.ai. Little fun fact here, it means builder in the Irish language, Gaelic. My name’s Patrick Aaron Murphy, a little Irish. The president of the company found that word togalai, and we said, “Holy cow it ends in -ai, this is perfect for us.” Togal.AI is the website, or shoot me a note, PEM@togal.ai and happy to give you a demo, get you onboarded, let you try it for a couple of weeks, and always open for comments, for ideas, for improvements. We have about 3000 users now in 12 countries, but we’re still nimble enough to make changes. We’re constantly trying to improve the software for users’ needs. It’s an exciting time in I think the industry, a lot of exciting new technologies coming out and appreciate the opportunity to talk with you. 

Steven Rossi: 

Absolutely. Yeah, it’s been a pleasure and I’m excited to see what you and Togal.AI do in the future. 

Patrick Murphy: 

Well, thank you. Have a great day.