How AI Is Automating Construction Estimating — And Why Atlantic Canada Trades Are in Demand Across Canada (Jeff Graham, Construction AI & Blueforce Logistics)

0:03Welcome back to the Atlantic Construction Podcast. I'm your host Daniel Arsenault. On this episode we have Jeff Graham of Construction AI and Blueforce Logistics. We'll be talking about artificial intelligence and construction and the role that it plays, Blueforce and its recruitment for construction workers, and more. Hope you enjoy. All right, welcome back to the Atlantic Construction Podcast. Very happy today to have with us Jeff Graham. Great to be here, Dan. Thanks, man. From Construction AI — that's me, yeah. And the AI in this case does stand for artificial intelligence.

0:37That's the idea. So we've got some really interesting topics today — we'll try and break them down and make them relevant to our listeners. So under that heading: machine learning, blockchain, and a bunch of other stuff. Happy to talk about all of it. Yeah. So to dive in, maybe just talk about how you ended up where you're at with this unique service in the construction industry. Well, it started in my childhood bedroom, Dan, as a grown man's dreams do.

1:13So what happened is I have a degree in communications from Simon Fraser, and I kind of worked in that field for quite a while. Along the way, a lot of people had encouraged me to explore more of the entrepreneurial thing. So I ended up doing estimating consulting with my dad, and that resulted in me being — I would have been in my early 30s — in my childhood bedroom, basically entering lines on a laptop.

1:42I was just sort of like thinking, there's got to be more to this journey, more to life. It's got to be — yeah, yeah. And that's, you know, it's not to disparage it — it was actually really important stuff. Like, we're doing civil estimating, it was a valuable service, it paid well, and all that kind of stuff. But I really struggled with, like, is this going to be my story? The tedious parts? Yeah, exactly. The tedious parts of it especially. And you know, given that I was kind of like —

2:09We did this for a few years, but because you're learning you kind of do the more tedious parts, right? Yeah, instead of the interesting stuff. And what happened is I had a kind of a chance encounter with a data scientist, and I told him what I did, and he's going like, "you know that can be totally automated, right?" And at that point I think we just had our fourth kid, and I'm sort of going like, I'm in trouble.

2:36The wheels started turning, right. It's like a little — yeah, a little bit worried about what I'd picked to do, in a lot of respects. And I mean, I guess the answer there is, like, the entire thing I think would be enormously difficult to automate, but there are certainly aspects of it that are very — turned out to be very realistic as far as automation goes. Okay, I think that's a good spot to ask this question: what's the difference between —

3:05— automation and machine learning? Yeah, sure. So machine learning is a type of automation. Automation in a way is like a function, right? So it's like those silly automatic dog food dispensers — my luck somebody who makes these things is going to be watching the podcast. You mean the silly automatic dog food dispenser? The dog food dispenser — I want to get one. So a function would be like: you load the dog food in the top, and the trap door opens —

3:38— and Bozo comes over and eats it, right. Like, that's a type of automation. What machine learning is — or a way of explaining it, I should say, because it is a very diverse field — one example would be: with machine learning you could give a neural network, say, 2,000 examples of, say, dog food, right, and it could look at any given bag of dog food and go, "oh, that is Kirkland, you know, large whatever," right, or Purina Puppy Chow —

4:14— so it can actually — the power is in its ability to predict something brand new, right, something it's never seen before. It can actually make a prediction on what it is. So yeah, the machine is learning. Yes, and how it's learning — this is why it's called machine learning! You're catching on so fast, Dan. I don't think you should cut that out.

4:54So we're having fun. But there are different ways that a machine can learn, and this is where I think people can get really tripped up. The type of machine learning that we do is called supervised learning, which is basically where you get a data scientist that is actively teaching the computer what to do. We're giving it examples, and basically, when it fails at its task, a data scientist looks at it and goes, "okay —"

5:25— why did it fail? Why didn't it correctly identify this part of the plans, or whatever?" And so they go in and they basically kind of essentially tweak the code or use different techniques to iron out these edge cases related to that function, right. Okay, so let me set up a scenario and you tell me if I'm wrong — just to make this relevant to, say, a contractor who has a quantity surveyor or an estimator in their estimating department, say —

5:56— three or four estimators in their estimating department. So they're going to benefit from machine learning in the sense that your software is linked with the estimating software that they're using, and a door count on a 12-story building — with anywhere from 30 to 50 man doors per floor — they don't have to manually click the mouse on all those door openings and count that data. That data can be recognized by the software as similar shapes and automatically — yeah. So we are really trying to get the computer —

6:35— to do the things that it's good at, and kind of leave the things that people are good at in that space, right. So with your example there, we can identify, say, all of the rectangles on a PDF, right — rectangles of the exact shape, or of an exact size — and then the computer goes, "okay, I found whatever, 37 of these rectangles — what do you want to call them?" And then a person —

7:02— looks at it and goes, "oh, those are doors." The computer, believe it or not, really struggles with going, "those are doors," because it sees maybe 8,000 rectangles — it could be squares, rectangles — and it's kind of like, I don't know, I don't know what the doors are. But we look at it and that's a really good classification task for us, right. This is kind of in terms of the earthworks thing — what our software can do is it can take a PDF of, like, a survey —

7:33— so with the spot elevations for an existing survey, and it can look at — it basically sees a series of, say, like 100, 200, 500 X's on a page, and next to those — topography, yeah, on a survey plan, right. And basically, next to those X's you'd have the elevation above sea level — could be like 62.81, and then that'll be 62.98, and then so on. And all of those elevations put together basically give you your topographical information, right. And as opposed to someone inputting those —

8:10— yeah, so this is a unique capability that we have. Basically we use machine learning to analyze those plans and we can extract an entire point cloud from the PDF for the estimator to put in the model. So whatever software you're using — it could be Agtek, Trimble Business Center, Kubla Cubed, whatever it is — we can take that data out and put it into your software and basically save you that real pain-in-the-butt task —

8:38— related to earthworks estimating. So it's like we can kind of group all these shapes together for you so that you don't have to count them. We also have something called a predictive layering function, so we can analyze a PDF and separate it out into its component parts. It basically looks at it and goes, "okay, I think these things belong together," and then often what it'll do is, like, "oh, that's all —"

9:06— your proposed road works in this layer," so we can take that and put that into a model. "Oh, those are kind of the engineering stamp — I don't need that, you can toss it out." But basically we make an attempt at separating this data into a hierarchy so that the estimator just saves a little bit of time. You know, some of that data would go into the existing model, some of it will go into the design —

9:32— model, some of it just isn't relevant to what they're bidding, or is just something that was on the PDF — like an engineering stamp, right. So essentially, Construction AI's software services are providing value mostly in the estimating field, or are there other areas of construction where that same value is being extrapolated by the individuals working as estimators — or as — yeah, I mean, so right now it would be primarily like pre-construction estimating, right. But it does have implications for —

10:05— other things, right. Like, we can read text and numbers, and it doesn't mean that there isn't more potential for it down the road — okay — as far as other construction management. Yeah, it's extremely interesting. So let's talk a little bit about Construction AI the company. There are a few partners involved, right, and these other partners are throughout the country, western Canada? Yeah, that's right. Yeah. So my business partner is Dr. Jason Heard — he's a researcher —

10:36— he teaches at Mount Royal. He's been in machine learning and robotics for — I mean, it's been like 20 years — but he's also one of those people that's kind of tinkered with computers his entire life. And then we have some investors, primarily out in western Canada. These would be owners of construction companies that have mainly been in my network, but they know the pain related to estimating and just how much work these things are. Yeah —

11:07— and you can show them where the value is going to be — so much appreciated in those monotonous daily tasks of counting, that's right, and tasks along those lines. Yeah. And these investors are strategic in that — so I guess the part of construction that I know enough to be dangerous on, at least, is earthworks. But one of them is in modular construction, the other one does residential homes, so we've kind of got a bit of that —

11:37— commercial, a bit of modular, a bit of residential — and they're kind of able to educate me on some of the knowledge gaps that we have. So we've got a good team of people that I think have some bench strength in terms of industry knowledge. Yeah, 100%. Let's just, for some context, where does Construction AI fit in in the grand scheme of things? From what I'm hearing, you guys are providing this automation, machine learning —

12:08— additive to the software, and you're partnering with large construction software companies that are providing estimating software, whether it's tender software, project management software, and you're teaming with them to provide a service that has these benefits. Yes, faster. So we have validated the technology with existing construction companies — we've had some earthworks companies that have used the software and basically validated that it does what it's supposed to do, which is obviously a really good first step. And then, yeah —

12:43— in terms of our commercialization model, the primary partnerships that we're pursuing are actually with the large players. These would be companies that already have a significant existing customer base, where we're essentially kind of a cog in the — you know — that monolith. Some of these are — one of the things I've noticed in the construction industry is there's been a fairly significant consolidation of a lot of these construction —

13:14— technology companies. Hexagon AB would be a big player out there, obviously Trimble, and that would be the type of company where, ideally, we're kind of slotting in. Okay. What is blockchain? Yeah, so it's a loaded question. I think a lot of people somewhat rightly associate it with cryptocurrency, right. The way cryptocurrencies work is that essentially it's a giant ledger of transactions and accounts. So it's like you've got — and all these things are just basically —

13:50— numbers, but essentially you have an ever-growing number of accounts and transactions and currency in this enormous thing called a blockchain. And the reason that it is so significant for so many industries — and I'd say construction and manufacturing especially — is that it's a type of database that can only ever grow. Like, you can't delete something off of a blockchain — it's just against its architecture. So it's an ever-growing record of whatever you've told it to keep —

14:28— track of. So an example for construction would be: with change orders, you could go, you know, the client changed the paint from red to blue, to yellow, to green. Historically, you might have deleted the existing colors. With blockchain, there would actually be a record of each time the color was changed. So it creates an audit trail going back on whatever you were — so it's a database architecture, and in a way the idea —

15:05— is really simple, right. I feel like it sounds like a fancy word, but it's a simple idea. It's just like you can't delete stuff now — you have to build your software in a way where it's growing, so account for that. But you're not overwriting stuff in a database anymore. Like, if you can picture this in your mind's eye, the way database architecture has been is like an Excel —

15:31— spreadsheet. And in that Excel you can add stuff, right, but you can also overwrite stuff. And once it's overwritten, it's gone. Whereas with blockchain, it doesn't allow you to do that — you can't delete things. Yeah, exactly. Or your device shuts off, or it's not saved to a hard drive — great example. Yeah. But with blockchain it's something that grows over time, right. And so you get this really robust audit log for whatever it is —

16:04— that you're doing. So it has really significant impacts for things like contracts, right. I mean, the whole thing with Ethereum — which is the second most valuable cryptocurrency in the world — is that it's a series of smart contracts, or kind of like relationships between different holders of this currency. And I would say that it has really interesting impacts for construction. Now, just to be clear, that isn't a space that I'm playing in per se, but I would say that it has —

16:33— it's something that I think is going to be emerging in the construction space in the coming years, just because it's such a great application for it. So is there any example currently where blockchain is being used in construction software, whether it be PM software or CAD or design software? Yeah, so again, it's not something where we're necessarily developing for the blockchain per se, so I'm not really too aware of exactly what might be on the product road map for different companies. But it would be more like — I think we can get a pretty strong sense that it will be —

17:04— a big part of what we have going forward, and those examples that you mentioned — I think it would be amazing for a project manager to get a record of all the different — what I mean —

17:32— gosh, there's so many examples, it's actually hard to just pick out a few because it's endless. Like, even — well, you mentioned change orders. Change orders, VMs — yeah, I mean, that's huge. Some jobs have hundreds. Yeah. Addendums on tender software, right, and then there'd be a series of sign-offs. So it's like a record of every sign-off, who signed off on that — like, all of these —

17:59— things that kind of build on each other. And it's like an ever-growing record — a construction project should be an ever-growing record of what the heck is going on, or what happened here. But it's one of those things where blockchain just gives such a robust log of activity. Yeah. Wow. I think it could also be useful in the estimating space too, right? I mean, or maybe more in the design? Yeah, why not — if you're an architecture —

18:36— firm and you're logging all your different designs from, you know, point A to point Z before the project is even finalized. Yeah, absolutely. So it's like — just think about a workflow, and it's like, who made that change to the design? With something like blockchain, it would all just be right there. It's also a way of storing files — there are different applications for it —

19:09— but essentially we develop our software on something called Amazon Web Services — they're the largest cloud provider in the world, and they're owned by Amazon. One of their newest offerings is something called a Quantum Ledger Database, and essentially it's a database where a series of files can be written to this architecture that just grows, right. So you could fairly easily — if you're a design firm — you could just start —

19:42— saving your files to something like one of these ledgers, and you'd have a record of all of this design change activity that's happened even before you got to the client. Yeah, very cool. So I know that undoubtedly people working in construction in Atlantic Canada — whether it be as quantity surveyors and estimators or project managers — are benefiting in some way from automation and machine learning in the software that they're using. For sure, that's obvious. But I'm just wondering, for Construction AI and yourself —

20:19— being a Nova Scotian, what your plans are with Construction AI locally, within the local market, and you mentioned you're planning on building a team here and that sort of thing. Yeah, so I think Nova Scotia is a very — and I would actually say, I'm just saying this because it's not from personal experience — Nova Scotians, from personal experience, Atlantic Canada would just be kind of more of an —

20:48— anecdotal experience — it's a really great place in terms of startups right now. Younger startups especially. There's a lot of really good government support through, like, Innovacorp, ECOA, there's the CBDC, there's — because we're so research-focused we leverage SR&ED tax credits, there's also NRC IRAP, which is huge. So in terms of our — I guess, sorry — I'm kind of interpreting your question as: why Nova Scotia? A big part of it is the support that's —

21:24— here. It would also be a really great place in terms of cost of living. And so much of this stuff now can be done remote — you're not necessarily expected to move to Menlo Park to do a startup nowadays, especially since the pandemic. And what makes Nova Scotia even more compelling is that you actually have a lot of really good researchers at the local universities. Like, StFX, Saint Mary's, and Dalhousie — they all have —

21:56— thriving data science components to their computer science programs, graduating with all kinds of wonderful degrees, like computer vision, and using time series data to predict events based on what's happened in the past. Yeah. So it sounds like there are a lot of aspects about the local market and the local scenario that feed into your business thriving in that regard. Yeah, I would say. And obviously a really good construction —

22:34— scene — that goes without saying. 100%. It's been neat for me because I'm not from here — I grew up in British Columbia but was born in Alberta. And it has been really cool being so welcomed. There's, in both the construction and in the startup scene, it's been really easy to get an open door and a conversation. And I think it's a bit of a Maritimer thing —

23:08— just the hospitality and the openness to new ideas. It's been awesome from that standpoint. Yeah, and it's so interesting to hear — excuse me — to hear you explain the different services and the different angles that Construction AI works from. For someone like me who had no prior knowledge of how this sort of stuff happens on the back end — but now you do have knowledge! Now I do —

23:38— and that's great. It's so interesting. And you know, you're doing those things throughout the day, and as the years go on, these monotonous, tedious tasks just disappear. And it's not like it just happened — there's actually so much going on behind the scenes. There's somebody that beat their head against the wall for years — hopefully not that long. Yeah, yeah. Like, man, when I look at just what you can do with a —

24:10— smartphone now, it gives you a huge appreciation for how complicated and how difficult a lot of this stuff would have been to research and build. But it's already in our lives, right. Like, when you talk to Alexa — there was a lot of research that went into your ability to basically order a pizza on your speaker. Yeah, right. There you go. It's like there's a lot more going on there —

24:44— than meets the eye. It's making decisions based on speech, and there are doctoral degrees related to figuring this stuff out. And then, even just being able to write text and have a computer understand essentially in natural language what you're saying — like "give me a pizza" — it used to be that computers were just so rigid. If you didn't enter the precise command, you wouldn't get the function that you wanted. So much of the research —

25:17— has actually gone into making these things so easy that, you know, my five-year-old could do it. Should deliver me — I don't know why that's such a good —

25:32So what are your hopes for the cutting edge — the tip of the sword — automation and machine learning in the software, in the construction industry? Is there anything that's really interesting that you're seeing down the line that could affect the daily tasks of people working in the commercial and residential construction industry in Atlantic Canada? Yeah. So I — it's a bit of a sentimental answer to this, right. But when I —

26:08— dream about a future related to this, I don't necessarily daydream about racks and racks of servers doing this amazing task that changes the world. I mean, it's a part of the vision, but what I actually think about is — construction is so hard, and it's so time-consuming, and there's so much that can go wrong. And what I really hope for with —

26:37— people is that they get a piece or two or ten of their life back. It's like, I hope that they get to take more vacations, I hope that they're less stressed about their job, I hope that it saves them time that they weren't going to have before. And maybe for people who are really ambitious, maybe it allows them to bid a job or achieve something that they wouldn't have been able to do otherwise. But just — I really —

27:05— hope that our research makes us a little bit more — I find that so much of how technology is designed, it sort of seems to separate us from our lives. Well, I think earlier — just to piggyback that — we'd used an analogy. You take credit for it — what was it again? We talked about how social media — social media, or a lot of technology, uses the way we're wired psychologically and uses it against us, and eats up our —

27:38— yeah, to present scenarios where we're going to become addicted and altering our behavior. Yes. And so what you're trying to do is, in the same space — specific to construction — leave the things that are good for humans to be doing, and take away the things that computers can do much faster. Yeah, yeah. That'd be — we don't want to do it, and we don't need to do it. Yeah, absolutely. And whether it's counting the —

28:11— number of doors and frames in a whole building, or calculating the elevations of a survey on a site plan, or any number of similar scenarios that might involve tedious, monotonous counting and entering data — etc. To put a button on that thought, Dan: with social media, it's kind of circling back to the description of machine learning. If you give a computer a bunch of examples of something, it can make a prediction, right. How it —

28:47— works with social media is — if you use Facebook, if you're a typical user, you're giving it thousands, tens of thousands, maybe hundreds of thousands, and perhaps millions of interactions with different things. So it'll go, "oh, this person really likes this type of cat," and its recommendations are based on feeding a machine learning algorithm all these examples of what you're interested in, and it gets better and better and better at serving up exactly what you want. And —

29:23— so it is very much using machine learning and AI to essentially distract you more. The reason that they want to do this is the more engaging those platforms are, obviously the more you stay on, the more advertising revenue gets generated. And that's ultimately what would serve the interests of the shareholders of these companies, many of which are obviously publicly traded. Whereas what you're doing with Construction AI is using machine learning and automation to take away tasks that we don't need to do —

29:55— that are monotonous. Yeah, that's right. And then hopefully more people will go fishing. It's like, or you can divert that time to something else that's productive at work — like negotiating a better deal with a supplier, or thinking about a way to plan a job that is going to make it more efficient, or to run smoother, to do better work, back-schedule. These are cost savings —

30:25— for business owners who have quantity surveyors — a number of them — where this automation is saving man-hours. So in my experience, estimators are among the most skilled people in a construction company. And so you're getting one of your top guns to basically look at a bunch of X's and go — it's like that person is enormously valuable and has just — yeah, he should be reading specs, or she — looking through specs? Yeah, or she — yeah, yeah, absolutely. Like, this —

31:02— is a person that very likely has an advanced engineering background. They can really help your business in so many ways. It's like: don't let them spend their time counting stuff. That's ridiculous. Yeah. But I don't want to get too much on my soapbox here. But yeah, that's in a nutshell what I'm trying to do. Amen. We should talk a little bit about the other thing you have going on — Blue Force — Blue —

31:33— Force. Yeah, sure. So there's an interesting story here. Maybe three years ago I was talking to one of the investors of my company in BC, and he goes, "man, I can't get people fast enough" — like, in the white-hot construction industry out there. Yeah. It's like, "I can't find people like that," and they're building modular homes — or something. That's right. Yeah. So he needs — it's like, "I'll take any warm body. Do you know anyone? Give me the bodies —"

32:02— anybody, anybody." And as it happened, my neighbor in Musquodoboit Harbour — this is somebody, his name is Wesley King — we kind of hit it off. And I was like, "what do you do?" And he's actually got two jobs — I'll tell you the other job in a second, it's kind of cool. But one of them is sending people out — to other places in Canada — it's primarily Maritimers going to other places in Canada to —

32:32— perform skilled trades work. So this has great application for building the guts of e-commerce systems — like conveyor belts and sorters and that kind of stuff — so that'd be a big vertical for Blueforce Logistics. And then another one would be agricultural manufacturing, so these are like welders, mechanics, millwrights. Yeah. And then lately laborers too — so not necessarily a CWB welder, but somebody that knows how to use one. Having basic safety training and being able to read a set of —

33:02— plans would be like key prerequisites for, like, conveyor assembly. Some of the tasks can get pretty complicated, and obviously with agricultural manufacturing that gets a little further up in terms of the skill level. So the name of the company is Blueforce Logistics Limited. Yeah. And so you guys are recruiting on behalf of the companies? Yeah, that's right. Yeah. And the —

33:26— owners of the company are really wonderful people, they've been really good to me. It's like, I think they're performing a really valuable service, right. Because where are most of them going — like, is it to a certain province, or was it initially Alberta and now it's all over? All over, from — yeah. So the company has its roots in the oil sector, but it's diversified: agricultural manufacturing in Saskatchewan is a big one. But —

33:57— e-commerce fulfillment is everywhere. I mean, it's like Amazon. And when you say that, you mean they're actually constructing these conveyor belts and sorters and those sorts of things? Sorters and conveyor belts, yeah. Caster decks is another thing, like for airports. But it's like all these things that are the backbone of how we basically order our dog food. The amount of work that would go into saying "Alexa, I need dog food" and then it shows up — there's somebody that had to make that conveyor. Dog food, brother —

34:36— pizzas and dog food, there was a lot of work that went into that entire process. Like the natural speech recognition that told the system — yeah — and you picked and packed it, and it went on a conveyor belt, which went on a truck. But it's a remarkable amount of innovation and work that would have happened over time to make that possible. Yeah, so true. So just for some context, how many east coast tradespeople in the run of a year have you placed — hundreds? —

35:04— yeah, from Blueforce, placing them? For whatever company — whether it be building modular homes, to welding agricultural equipment, to e-commerce — yeah, absolutely. Hundreds. So how do you find these tradespeople? Or do they find you? Both, yes. So Indeed is a big source of applicants. Facebook is another one. Interesting thing about Facebook: it's enormously — in the Maritimes, it's popular everywhere, but it has a particularly high penetration out in Nova Scotia, PEI, Newfoundland —

35:39— and New Brunswick. So that's a huge source of inbound. Yeah. Facebook ads, and then organic search would be another one, and word of mouth — which is going to be a great marketing tool in the east. Yeah, absolutely. So let's say, for example, I was a carpenter and I engaged with Blueforce — I found a job in — so we would find you the job. You find the job? Yeah. And then you take care of their flights, you take care of everything it takes to get that person working on the job site? Yeah, so the —

36:08— value proposition — I'll go through it quickly — the value proposition for the worker and for our clients. For the worker, it's basically like: "hey, I'm a CWB welder and I have WHMIS and this type of safety training." We go: "okay, cool, we'll see what we can find for you." Lo and behold, there's an agricultural manufacturer in Saskatchewan that needs 10 people, or whatever, and this welder fits the bill. What we'll do for the welder is we will set them up with a flight, a car, accommodations — everything that they need to get on site. So they —

36:37— are not out of pocket a dime, right. They just basically hop on a plane, go out, do the work. It's often like, say, an eight-week rotation —

37:06— sometimes these projects can last for multiple years, but it'd be like eight weeks, then you go back home for a bit, kind of unwind, and then go back to work, and that cycle continues. And then for the client companies it would be like: "hey, we've been trying for three months to get this type of person — I don't know, say a CNC machinist for a night shift, and it's just not gonna happen." So that would be when they would call —

37:34— Blueforce and say, "we need this person, we can't find them." Or another one would be: "hey, we bid three jobs, we thought we'd get one, but we got all three, so now we need to ramp up our crews enormously for a relatively short period of time and then go back down." Like, you're not comfortable necessarily with taking on all of that overhead permanently. So can Blueforce place a Maritimer with an east coast company? Oh yeah, they wanted — yeah, that happens. It's not —

38:02— like you're always sending people to other western provinces — you're placing people here too? Absolutely, yeah. No, it absolutely happens. I guess I'm just speaking more in terms of how it generally works, which is that you have a lot of people from the Maritimes who go out west for work. We do take people from Alberta and BC and move them as well — it's just that for whatever reason there's an abundance of skilled trades in the —

38:28— Maritimes, and I think it's a bit of an underappreciated thing. We've got lots of really good people here, and it's not like a lot of them have to go somewhere else to find work right now, being as busy as it is in construction. But maybe there are certain trades where that's necessary — it really completely depends, right. Obviously construction is really booming right now, so maybe some people that would have been looking for work elsewhere — it's less —

38:54— necessary. But yeah. So I wanted to ask about Blueforce — who are your competitors? Are there any other recruitment firms? Yeah, but specific to construction — well, it would be like — there are very few companies that do exactly what Blueforce does. But then there are many companies like Matrix, so the key difference between Blueforce and these other companies is that with the other companies they're generally restricted to whatever the local labour market has. So it's like if —

39:33— you can't find a CNC machinist — and you're using that example because it's a rarity, it is a night shift — you've got to find the right person out of a thousand that's going to be right for that job. Yeah, and if you can't find them locally, why would a temp agency be able to find them locally? Would be my question. Yeah. With Blueforce, the value proposition is that we will draw from a much larger —

39:58— talent pool. We'll scour every square inch of Canada to find you the right person and bring them on site at a reasonable cost. And you're placing hundreds of Maritimers in trades positions — not just Maritimers, but a huge chunk of them are Maritimers, right? Yeah, and we're placing them all over. And that stems from the trend of the modular homes in and around Alberta where the workforce was needed — so that was a trend at that time, and now there could be different —

40:28— trends. Yeah. And obviously it's a never-static thing — you kind of adapt to what's going on in the world. It's an ever-changing thing for sure. Very cool, man. So for some of our listeners and our audience today, if they want to find Construction AI and know more about the software and the machine learning and automation and all the services that you guys provide, and also Blueforce, where can they —

40:58— where can they find you online, on social media? Well, blueforcelogistics.com is for Blueforce, construction.ai is for Construction AI. But honestly, I would love it if people connected with me on LinkedIn. My name is Jeff Graham — if you type in "Jeff Graham Blueforce" or "Jeff Graham Construction AI" you're going to find me. I'd love to have listeners connect with me that way and be happy to answer questions or want to —

41:28— chat or whatever. I'm always looking to make new connections, so that would be my preferred way of people reaching out to me. Awesome. Unless you think I'm a big jerk talking about dog food all the time. Never — those two topics are right down my alley! Especially with the extremely interesting topics that we've had today, it's been a pleasure. Seriously, some really interesting stuff. And it's not every day you get a chance to —

41:55— talk about that side of the industry. Yeah. With Blueforce and the recruitment and those situations, but especially with Construction AI, it's neat to get a glance at how that stuff happens behind the scenes when you're someone who's involved in construction on the other end. And it's so valuable to us and saves us so much time. It's great to contrast that with someone like yourself who's in it and immersed in that side of the business. So it's —

42:23— it's been amazing. Yeah, I really appreciate it, Dan. Thanks for having me on — it was a lot of fun. Cool. Thanks for tuning in to this episode of the Atlantic Construction Podcast. Be sure to follow us on any podcast platform you use. You can also find us on LinkedIn and Instagram: Atlantic Construction Podcast. Be sure to send us a comment or a review — we'd love to engage with you.