What do you get when you combine 3D cameras mounted on construction workers’ hard hats + visual recognition technology + artificial intelligence? You get OpenSpace, a fascinating startup that’s using new technologies to reduce costs and inefficiencies in the construction industry. OpenSpace enables builders to track progress remotely, without having to visit construction sites to examine every little detail. It was a critical tool during the pandemic lockdowns and adoption has only kept growing since.
Here’s Jeevan Kalanithi on Crafted, Artium’s new podcast.
Listen to all episodes and subscribe here!
Jeevan Kalanithi: Once COVID hit, it was acutely painful. It's like we cannot have people on these projects because they may make each other sick. We can't travel. How the heck are we going to do our jobs? And they were like, Hey, there's this thing called open space now. We don't have to fly the architect out. He or she can just log in.
Dan Blumberg: That's Jeevan Kalanithi, Co-Founder and CEO of OpenSpace. OpenSpace allows builders to see progress without having to personally inspect everything. What's the status of the seventh floor? Are we ready to pour concrete? Is the rebar installed properly? At an OpenSpace, construction site workers wear 3D cameras on their hardhats and they just go about their jobs like they ordinarily would. Meanwhile, open Space uses AI to create Google Street View style maps of the sites and not just from today. You can also go back in time to see behind walls and under floors. It's been a game changing technology for the construction industry, especially when COVID turned our world virtual and inspectors couldn't safely go on site. Jeevan joins us today to walk us through the product.
Jeevan Kalanithi: That you can kind of teleport into a space and see what's there without leading to physically be there.
Dan Blumberg: The problems they had to solve along the way.
Jeevan Kalanithi: It was like, okay, yeah, it's definitely a cool idea but there's no way in hell we're ever going to use it.
Dan Blumberg: And how they eventually nailed the user experience. Welcome to Crafted, a show about great products and the people who make them. I'm your host, Dan Blumberg. I'm a product and engagement lead at RDM where my colleagues and I help companies build fantastic software and recruit dynamic teams. I want to start with MIT Media Lab, where you're a graduate student because it seems like it was a formative two years for you that really led to a wide variety of startups that you've launched and products that you've worked on. From games to drones to 3D cameras on construction workers' heads like you're working on today. Can you tell us a bit about MIT Media Lab? What drew you to it and the kind of seeds that it planted?
Jeevan Kalanithi: Yeah, I'm happy to talk about the Media Lab. It's one of these very special places on the planet and I really treasure my time there and it was hard to leave actually because it's so fun. It's a couple things I would pull out in terms of the things you learn there, some of which are fabulous lessons for getting companies started, some of which are things you must massively unlearn, actually. The Media Lab collects people of a variety of skill sets, most of which have pretty deep technology background, but some have design skills as well, or some are artists. And the goal is to build systems that are based on technology that you believe would improve or transform the human experience in some way.
So you learn a lot about how to build things that you can figure out if it will work and demonstrate it quickly, how to prototype things well. And you also learn about how to communicate those ideas in a way that people will get quickly and they can get excited about. And of course you meet extremely brilliant people. I would say things to... Had to unlearn are things that people think are neat or cool or might be successful TED talks, those are not always great businesses and so a lot of my experience post MIT has been learning the distinction between the two.
Dan Blumberg: Were you actively looking to found a company or were you just kind of having fun or could you not tell the difference sometimes?
Jeevan Kalanithi: I definitely was not actively looking to start a company. I went there because it just seemed like an interesting place to be and I had a variety of interests prior to the Media Lab that seemed very disparate and to me, problematically disparate. I found myself working on some technology projects and writing software for certain things or working in a research lab but also working on interactive art projects.
Dan Blumberg: And so connect the dots, how so 10-ish years later, you're founding a company in construction technology. Not sure if you saw that coming. How did you get there?
Jeevan Kalanithi: I'd love to tell you again that it was like this highly irrational analytical process and I know people like that and I admire them, but that's not really me. A friend of mine and I were working on a project at the Media Lab that people were very interested in and we just were invited to do a talk at the TED Conference. We were like, "Okay, sure, that sounds good." We have three months to get this thing to work so that we can give the talk, which we did do and it went great. As one of those talks that a lot of people reviewed on the interwebs and so on.
Dan Blumberg: David Merrill presented the project, an interactive gaming system at Ted.
David Merrill: Spatial reasoning is deeply connected to how we understand a lot of the world around us. So as a computer scientist inspired by this utility of our interactions with physical objects, along with my advisor Patty and my collaborator Jeevan Kalanithi, I started to wonder what if when we used a computer instead of having this one mouse cursor that was like a digital fingertip moving around a flat desktop, what if we could reach in with both hands and grasp information, physically arranging it the way we wanted.
Dan Blumberg: Hardware plus software challenges really excite Jeevan.
Jeevan Kalanithi: I think the ideas that I was attracted to have always been those that are experiences for people that allow them to exist in real physical reality and get the benefit of what software and digital systems can do. I tended to not be interested in things that are purely in the software ethereum. I think the ideas I've worked on have always had that very tangible kind of hardware component.
Dan Blumberg: The TED Talk was a hit.
Jeevan Kalanithi: And we thought that there could be maybe a business idea in there and people were telling us that there could be. So we said, "Let's do it. Let's start a company."
Dan Blumberg: Sifteo was a venture back to gaming startup that Jeevan would go on to sell to 3D Robotics in 2014 and that planted the seeds for what is now OpenSpace. At 3D Robotics, Jeevan worked with drones to solve problems for clients in construction. This led Jeevan to an aha moment. He was working with drones and unmanned aerial vehicle systems for builders.
Jeevan Kalanithi: It was really clear just spending time on projects that just knowing what was going on was hard and that a picture really is worth a thousand words. The drone system just took a bunch of pictures from the top down and provided a record of what was going on.
Dan Blumberg: The drones were useful, but they could also be difficult and didn't paint the full picture.
Jeevan Kalanithi: The drone solution was cool, but it was definitely a technology looking for a problem and it found a problem, but it wasn't really built for how those folks live their lives. Drones are cool, but they're a bit of a pain to use, you have to get it out of the case and fly it around and not crash it to anything and there's regulatory concerns and more important they don't go inside.
Dan Blumberg: Turns out this was a pretty key issue to solve, but it wasn't until about six months later that Jeevan and his co-founders is decided it to replace the drone with a mobile camera.
Jeevan Kalanithi: What OpenSpace does is we make it extremely easy to have a complete visual record of really any space, indoors or out. And if you've used Google Street View, you can get a sense of what OpenSpace looks like. It's this fully 360 look around at anything on your phone or on your computer so you can kind of teleport into a space and see what's there without leading to physically be there. Couple added things. We've made it very easy to not just image a space once, but over and over and over again, so you can measure change over time. That's really important in general. But if you've done a home remodel, you might want to know what's behind the wall. Right now, you have to rip it open and that is very expensive and annoying with our system. You can just rewind time to go see what was there before the walls were put up.
Dan Blumberg: OpenSpace captures these images with a 3D camera mounted on a hard hat. A worker throws on their hard hat and walks the site going about their work as normal. The camera then passively captures the video and the software utilizes artificial intelligence to create a virtual map of the space. Putting the camera on the worker's hard hat was a vital decision and it came from prototyping and listening to early customer feedback.
Jeevan Kalanithi: The key realization for us is I believe that good products should remove work, not add it, and good products typically do not ask their users to change their behavior. And the very first version of OpenSpace actually was not hard hat based. It was a camera on a tripod and you'd move it from point A to point B to point C to point D, and we did a bunch of slick things to make that easier. But fundamentally, you were still moving this piece of hardware around.
Dan Blumberg: Jeevan visited construction sites a lot. One day workers were pouring concrete. It's a critical step and one that you can't exactly hit the undue button on. Before concrete is poured, the rebar, the reinforcing steel is still visible. Jeevan realized this would be a perfect opportunity to document before and after of the concrete pour. So he took out his camera and while the results of the data were great, there was one huge deal breaker.
Jeevan Kalanithi: I was showing the output to a guy named Nick Merovich who was the PM who happened to be there that day. And, "Hey Nick, here's the latest and greatest, check it out." He was like, "Wow, this is pretty useful." And he was giving me lots of positive commentaries, so kind of glimmers of product market fit goodness. And then he said, "Hey, Jeevan, one question real fast, how long did it take you to take all these photos?" And it took a while. It was maybe 10 or 15,000 square foot floor plate. I'm kind of dragging this tripod from point A to point B. And I told him how long it took and he was like, "Okay, yeah, it's definitely a cool idea but there's no way in hell we're ever going to use it." And I was like, "Okay, good to know." And he was like, "Come on man. I mean you're here every day. Who besides you would have time to do what you just did? I mean, I'd love to have access to this data, but I don't know who I would ask to do it."
And I was like, "Okay, yeah, totally, I get it. You have a point. I mean I should have realized that myself." And then he said, "You know what would be good though? If I walk the project, it's part of my job to do it and my assistant, superintendents do it, so could I take that little camera of yours and just wear it or something? Like put it in my hard hat and I just walk around and then you guys do whatever you do, computer stuff and then you give me the same thing I'm looking at now, but all I have to do is walk around because if you did that, we would buy it." I said, "Okay, great." Marching orders received. And so that's where we really embraced that mode of capture. And it's a hard technology problem too, which was good because we were good at solving those. So we felt that would give us competitive advantage as well.
Dan Blumberg: Why was it a hard technology problem?
Jeevan Kalanithi: It's described in the literature in different ways, but oftentimes it's called indoor navigation in a GPS denied environment. What that means is you're trying to figure out where a thing is in the world or say reference to a floor plan, but there's no signal telling you where exactly you are. Let's say the user even tells the system they're in the lobby. Okay, so where they're starting then they just are walking around, whatever they're doing, they're going in circles or doing all kinds of stuff and all you have is the video file. Did they go right? Did they go left? Did they go stop? Did they turn around? That's all the information you have is a frames of video. It's very easy for a system to get lost and screwed up trying to figure that out. That's the problem we had to solve, is to kind of reconstruct the path of the person walking around with no input other than the video.
Dan Blumberg: I have a vision of you and your team walking in circles and running tests and are we getting it right? And is that some of what it looked like? How did you approach the problem?
Jeevan Kalanithi: You're exactly right. You'd just be walking around in loops in your house to your office and then processing it and then looking at the track and if it got it right or not, then making tweaks and a lot of iteration.
Dan Blumberg: As you went through that process of improvement, I imagine there were lots and lots and lots of 10% improvements when you adjusted this or that. Were there any moments where there was a 10x improvement because you unlocked something that you figured something out?
Jeevan Kalanithi: Yeah, definitely, definitely. Although maybe I'll zoom out to explain our approach to the design of the user experience and how it tied to the develop this hard technology. So Nick tells us this thing, "Hey, if you built this experience, we would buy it." All right. The first thing we needed to validate to see if this as a business is get someone to actually pay us for it. Because if they don't, then there's probably some issue with how we're describing it. The problem isn't real, whatever, right? And the technology problem is very complicated. So we were like, well, how can we answer this question on product market fit as fast as possible and not block on the technology working?
And so we said, well, what we could do is we could just have a system that was a Wizard of Oz system where people walk around that we get the video and then we just have us as human beings like reconstruct the path by just looking at each frame of video and manually pinning it to a floor plan and then ship the answer back to the customer. And of course you couldn't really do that at scale because it'd be so expensive and take forever, but we could certainly answer the question of would people buy this or not. And that's exactly what we did and we built a system that kind of worked in that way and people paid for it and they're like, "This is good." Okay. Good to know. Data received. Now it makes sense to really invest in the computer vision technology in a hardcore way to get it to work and be scalable, which of course we've done.
Dan Blumberg: I love a good Wizard of Oz or Concierge MVP.
Jeevan Kalanithi: Yeah.
Dan Blumberg: Absolutely. Don't invest the time if there's no there there anyway.
Jeevan Kalanithi: It's a common mistake. I would say product engineering is for technology companies, it's your most expensive lever and it takes the longest for it to get results, but it's the one that gives you highest leverage. So good to know. But if you can answer questions without invoking the need to use it, you should absolutely do that as much as humanly possible so you can learn quickly and then only say, okay, we're going to release the engineering beasts to make this really work until you're really confident that it's something worth building.
Dan Blumberg: OpenSpace is now used by a diverse group of general contractors and builders. The platform has tens of thousands of projects and the data they've collected recently hit 11 billion square feet. Part of OpenSpace's rapid growth was being the right product at the right time.
Jeevan Kalanithi: It is a bit morbid to say, but is absolutely true that COVID was a huge accelerator for the company. I think there are a class of companies that experienced this. Zoom is one of them. I think there are some article that came up about OpenSpace that called us the Zoom for the construction industry. Pre COVID people definitely saw the benefit of having this data resource where they could look and see a source of truth for their projects. Once COVID hit, it was acutely painful. It's like we cannot have people on these projects because they may make each other sick. We can't travel. How the heck are we going to do our jobs? And they were like, "Hey, there's this thing called OpenSpace. Now, we don't have to fly the architect out. He or she can just log in." And so we saw a huge uptick in adoption and that was a big, big accelerator for the company.
And what's question I had is as we're coming out of COVID, would people still want what we provided to the market? And I'm happy to say the answer is yes. It's just a thing where once you use it, you're like, why would we go back to the old way? This is just better. Whenever our older customers, a superintendent had this funny comparison, it's like, "It's like I used to go to Blockbuster and then there's Netflix. Why would you go back to Blockbuster?" Like OpenSpace is the same thing. Why would I have to go back to having to get in my truck to go to the site for every little last issue?
Dan Blumberg: Fast forward a few years, where is OpenSpace going?
Jeevan Kalanithi: A big picture, I really want the users and customers we have to live in a world where if they want to know about the buildings that they're buying or using or building that they can just rely on the fact that those buildings have a visual memory of themselves. It's like the leap from having to physically go to the library to look information up. If you wanted something, just Googling it. And I would take it for granted now, but people that work in physical reality can't just Google it. So that's thing number one. Number two is to help them answer questions that may take a while by simply looking at the images that if we can process the data and simplify it, then we want to do that too. So progress tracking is a good example of that.
Dan Blumberg: Progress tracking is traditionally a very analog process where someone walks the site and checks off what's been done. Open space offers a core capture product where that's done virtually and automatically and there are more problems Jeevan wants to solve.
Jeevan Kalanithi: Maintaining and building buildings in the physical environment is just too freaking expensive and complicated. There's got to be a way to make that just easier and more efficient and reduce the costs. And that's something I'm really excited about too because one of the reasons why building and maintaining buildings is so expensive is that's just really risky and there's no way to quantify the risk easily, but the data we provide makes it easier. An insurance company can feel more confident about underwriting a project if they can see what's there and evaluate the builder or just deal with a claim more easily than sending people out. So we've recently started to announce some partnerships with insurance companies that basically say, Hey, if you're using OpenSpace, you get a preferred rate on your insurance.
And insurance is a very, very big chunk of the cost of a building. And it's just one example of the way where lack of data creates overhead and waste. And so it kind of seems weird that like a software company could make it cheaper to build a building. You think like it's got to be a material innovation or something. And I think those will happen too, but I hope that we can make a little dent there as well so that this housing project now pencils, people will build it and they might not have done it before if we hadn't existed.
Dan Blumberg: That's awesome. That's a great place to end. Thank you so much for your time. It's really, really fascinating stuff you guys are building.
Jeevan Kalanithi: Yeah, I appreciate it. Thank you for your time as well.
Dan Blumberg: That's Jeevan Kalanithi, Co-Founder and CEO of OpenSpace. And this is rafted from RDM. At RDM, we build incredible products, recruit high performing teams, and help you achieve the culture of craft you need to build great software long after we're gone. We artisans love partnering with creative people to build their visions of the future. If you've got an opportunity you'd like to discuss or just want to learn more about us, check us out at thisisrdm.com or drop us a line at email@example.com. This podcast is new and we'd love your support. If you like today's episode and hey, you've made it this far, maybe text a few craft minded friends a link to the show and please subscribe and join us as we highlight more great products and the people who make them. I'm Dan Blumberg, this is Crafted. See you next time.