Want to join the conversation? Connect with us on Tribe.
Find “Founders in Arms” on Apple Podcasts, Spotify, and YouTube.
Transcript of our conversation with Jeevan Kalanithi:
Raj (00:00.078)
Welcome to the Founders of Enormous Podcast. I'm Raj Suri, co-founder of Lima and Tribe. And today with me, have Jeevan Kalanidhi, who is the co-founder and CEO of OpenSpace AI. Is that right?
Jeevan (00:07.991)
That's more important.
Jeevan (00:29.619)
Yes, that is right.
Raj (00:31.154)
Absolutely. Yeah, so Jeevan, welcome to the show. First of all, let our audience know that Immad couldn't make it today. He is otherwise occupied. He'll explain what he's up to later. But today we have a great conversation talking about AI and Jeevan, you and I have known each other for a long time. When did we first meet? We met at MIT, right? Back in 2009. Yeah.
Jeevan (00:54.471)
At MIT, I feel like I met you at Anna's or somewhere around there. I don't know what we were talking about, but I do remember that.
Raj (01:04.163)
Yeah.
I think I remember you were interviewing at some point, you and David were interviewing to be consultants on Elac Heart or something like that. You were like, not interviewing exactly, but we were having conversations to be like a consultant, because you had a consulting thing for a while or you were wanting to be consultants or something, Yeah.
Jeevan (01:18.731)
Yeah.
Jeevan (01:22.563)
that's right. Yeah. Right after I finished at MIT, Dave and a guy named Brent Fitzgerald, we had a little consulting company called Taco Lab where we did some tech projects. They were all over the place. Remember, we consulted on a smart greeting card for Hallmark. then we actually very randomly, we did Sequoia's website, Sequoia Capital's website. That was like, it was two of us. And honestly, I knew like,
Raj (01:33.227)
Right.
Raj (01:40.501)
Mm-hmm.
Raj (01:46.242)
Really?
Yeah.
Jeevan (01:51.337)
All my software experience was either in AI systems or microcontrollers. Like I didn't know anything about web stuff. So branch really did all the work there. But yeah, we had a little consulting company in that, that we did before we started our day. And I started our first, first startup and we sold that one and then number two.
Raj (01:55.652)
Mm-hmm.
Raj (02:07.822)
Right. So you were at MIT Media Lab, Yeah, Media Lab. Was that a PhD or a master's? What were you doing there?
Jeevan (02:11.637)
Yep. Yeah.
Jeevan (02:16.278)
I have a lowly master's degree, not get a PhD.
Raj (02:18.734)
Yeah, nothing in Media Lab is lonely. So I used to like trawl the halls of a Media Lab trying to recruit, you know, people to come work on my stuff. And because I was so inspired by the Media Lab, you know, I was like, you know, these people are this kind of this intersection between like creativity and art and technology. And it's exactly the type of thing that inspires me. you know, I'm sure that, you know, you got a chance to like meet a lot of inspirational people. David was also at Media Lab.
Jeevan (02:22.583)
you
Jeevan (02:48.287)
Yeah, yeah. So Dave Merrill was co-founder of our first company, still a great friend, started an amazing company called Elroy Air, which is unreal. It's like a robotic airplane. And Dave and I actually know each other from undergrad. We were like friends, roommates, same major, in a band together, in fact. And then he went to MIT. Pretty much right after undergrad, I moved to New York and did a bunch of random stuff before rejoining him there and working on some projects together.
Raj (03:08.248)
Mm-hmm.
Raj (03:17.358)
And your first company was Sifteo. Is that right? Yeah, explain what this was. This was a really cool idea.
Jeevan (03:20.297)
Yeah, yeah, that's
Yeah, so I look back on that company with such fondness because it was such an interesting, crazy idea. essentially we built like a tangible game system, you know, for kids and families and it consisted of little intelligent blocks with displays that were aware of each other that people could write games for. And we did some cool stuff. We had like licensed Ninja Turtles game and had the creator of Magic the Gathering create something for us. We shipped the product into
You know, in the stores, had been two generations of it. You know, could buy it at Best Buy or whatever. So that was really cool. And that was based on some work that we did at the Media Lab. you know, being in right place at the right time is cool. We just got an opportunity to do a talk at the TED conference on that work. And we were like, all right, well, let's just do it. Why not? know, and Dave did an excellent job and that kind of sparked us starting that company and, you know, raising money and doing all
Raj (04:15.692)
Mm-hmm. Mm-hmm.
Jeevan (04:22.231)
which we eventually sold to a company called 3D Robotics.
Raj (04:25.361)
Right, 3D Robotics, that's Chris Anderson's company? Yeah, yeah, And he was the TED guy, right? He actually ran TED? Or, no, no, they use Wired. Oh, right, right. Yes. Okay.
Jeevan (04:28.406)
Yeah, yeah.
Jeevan (04:33.495)
There are separate Chris Andersons. It's very confusing. There's a Chris Anderson, that's the Ted Chris Anderson. And then there's a Chris Anderson that was the editor in chief of Wired and then went on to start 3D robotics.
Raj (04:42.37)
Mm-hmm.
Right, right, right, right. Yes. Yeah, I remember the product was called Siftables, It was just like a high-tech toy for kids. The type of thing that I wish there was more of, know, as a parent now, I'm always looking for things like that. And there's a lot of them, but there's, you know, I wish there were even more because there are a lot of fun for the kids to play with. And have you, you have kids, right? Have you had your kids play with something like Siftables?
Jeevan (04:50.261)
Yeah, yeah, yeah.
Jeevan (05:13.303)
Yeah, you know, it's funny, like, um, you know, I had my first kid, my daughter, um, while-
And then I've got two kids now, they're 13 and 10. And, you know, they weren't old enough to really, really play with the product until after it sold the company. and, but, you know, I have, you can still get them. mean, like they're not manufactured anymore or anything, but I have some of the siftables, right? Sifty O cubes is what we call them after we, we turned into a company. I'm a shelf there and the kids just.
played with them one day when they're old enough and it was cool. That was amazing. And they just sat there and played with them for a long time. And I was like, all right, that passes the test. mean, that's ultimately like, if they like and they use it, then you've done at least part of what you're supposed to do. You're starting a company, have users who like the thing that you built.
Raj (06:08.482)
Yeah. Exactly. mean, my model for building anything in consumer is like, A, you have to like yourself, B, your friends and family have to like it, and then C, like people who are their friends and family have to like it. Like you have to go to like increasingly outer layers of, yeah.
Jeevan (06:24.151)
Exactly, it's like many things in startups. It's like first prove it to yourself that you can do it or like it. But it's not enough. That's step one. Then can you get your outer ring? And then if you can get people that are basically strangers to be on board, then you might be onto something.
Raj (06:40.822)
Exactly. Yeah, that's how I think about it. mean, me, Siftable and Siftio is very unique because it's hardware, right? And then that's like one of the hardest categories to innovate in. I mean, if you look at the company, there's so many companies that have actually not done well in that space, but it feels like it's a, mean, when you succeed, it can be a massive hit, right? Like there's obviously the iPhone is a consumer hardware product and...
Jeevan (06:56.501)
Yeah.
Raj (07:05.374)
stuff many companies have gone public fitbit and you know, gopro even though they actually, you know, they they tend to hit a peak at some point but you know, How do you feel about how that category is evolved because you almost see no companies like this getting funded These you know in the last 10 years basically
Jeevan (07:18.699)
Yeah. I think it's super, it is super difficult. mean, yeah, we ran the company and we sold it and that's good, but the actual frequency of truly phenomenal outcomes are very low. and, and they consume a lot of cash. mean, it's, I think it's very hard because the way I would say it is like,
you know, there are consumer hardware products, and then you have to say, which ones could be, most likely to be developed by a startup effectively, and which are most likely to be by large company. Truth is, large companies generally have most of the chips stacked in their favor for consumer hardware in a way that's different than other categories. You need a big capital base, the channel's very important.
And oftentimes, I think in consumer hardware, you can be a victim of your own success because if you have something that works well and people like it, you face these two challenges. One is the more products you have to make, the more cash it consumes, they're very capital intensive. And yes, you can get debt and things like that to do that, but it's never quite that simple. A person who was on my board at...
Raj (08:32.014)
Mm-hmm.
Jeevan (08:42.229)
Siftio and is on my board now actually, awesome guy named Alan Hendrix. He was multiple times CFO to companies public, really one of these gold Midas Touch guys and he's awesome. I've learned a lot from him. He was the CFO of Pure Digital. Remember the flip cameras and that company did well and Cisco bought them for a bunch of money. And he always said, you know, every year you're betting your company because you have this one window Q4 to see if people like what you have or not.
Raj (08:56.448)
Mm-hmm, for sure.
Yeah.
Jeevan (09:11.123)
It will, it will type all your cash to, to fund the product. And you don't even really know if people like it until you ship it. So the, the loop is very open between when you have the idea and whether you get actual feedback from the customers. and you can't absorb a mess and the larger companies could, you know, they can ship a product that that is,
Raj (09:28.142)
Mm-hmm.
Raj (09:34.798)
Mm-hmm.
Jeevan (09:36.703)
not a winner and they won't get killed. I Apple has shipped stuff that is not good, but people don't like, you know? But if you do that as a consumer art company, you can die. And the other thing is that if you create a category, the R &D cost to develop Gen 2 and Gen 3 and Gen 4, they oftentimes go up because you have to ship more more sophisticated features that are just expensive to develop.
Raj (09:57.303)
Mm-hmm.
Jeevan (10:04.503)
And that's pretty gnarly from an OpEx point of view. yeah, it's hard. mean, whatever. Any startup is hard. They're all hard. So yeah. Yeah. Yeah.
Raj (10:07.191)
Yeah.
Raj (10:14.35)
Yeah, they all have different types of hard, I agree, but there's some startups which are easier than others, I think. like SaaS companies in general are easier, I think, than consumer hardware, right? Because you can deterministically find a formula for success in the SaaS company. Your customers give you very regular feedback. They'll tell you exactly what they want, right? And you can kind of iterate your way to success. As you mentioned, I think...
Jeevan (10:23.009)
Yeah. Yeah. Yeah.
Raj (10:43.022)
this consumer hardware thing, you need a little bit of luck probably early on and then you also probably need to fund the company well. then you need to be really skillful as you said in Gen 2 and 3 as you grow the category. But for me, I'm fascinated by that because I think some of the biggest winners in the space and that dominate our lives today came from consumer hardware. Tesla is a good example of that.
Very difficult company to build right, but it's you know, it when it when it succeeds it's you know, it's a dominant company, right and And and you know over trillion dollar. I think they're a trillion or the closer trillion in market cap at this point
Jeevan (11:18.487)
no, totally.
Jeevan (11:24.341)
Yeah, yeah. But I must be companies like that as like outliers. They're not necessarily object lessons for other founders because, you know, if you don't have the credibility to raise absolutely insane amounts of money,
Raj (11:33.454)
Mm.
Jeevan (11:44.695)
You just can't like if someone says, I want to start the next Tesla and there aren't already like very wealthy and can pull in massive amounts of money. I would be like, okay, maybe maybe do Tesla your version of Tesla after you've, you know, founded and taken up software company public, you know, and you have that pull to
Raj (12:05.934)
Yeah, I agree with that. I agree with that. A first-time founder should not be trying to start Tesla. Yeah, it's um, you need to have you need to have had a win I think in your career Um, so tell me a little bit about what you're doing at open space. Um, I would I think I talked to you right before you started this company where you were um, You know incubating it you incubated this out of lux capital. Is that right?
Jeevan (12:10.741)
Yeah.
Yeah.
Jeevan (12:25.143)
Yeah, yeah. after my tour at 3D Robotics, I took some time off and didn't do anything, which was great. And then I started working on OpenSpace and started it with a couple other MIT Media Lab buddies, Mike Fleischman and Philip DeCamp. We both did their PhDs. It'll be an AI and computer vision.
Jeevan (12:48.531)
Yeah, I'll give you the nutshell of what we do and I'm happy to say like why we bought the start the company, but it's a pretty simple idea. Like our mission is to simplify how the world gets built. Like we want to build computer vision and AI power tools, where people actually build real stuff, build real buildings, like construction people.
And the core product is pretty straightforward. We've made it really easy to have a 100 % complete visual record of any building inside and out. Kind of thinking about a Google Street View of a project, if you could just be building a building. If you've done a home remodel, you could probably even imagine the value of this. You just have this complete record of everything, and you have, therefore, the instant replay, time machine, truth serum, call it whatever you want. There's a way to see what is actually going on from anywhere at any time.
basic idea and we sell it to builders. And it's going pretty well. We've amassed a large amount of data, or customers have collected, think, 40 billion square feet of imagery data. And give you a sense, the Empire State Building is like three million square feet, so that's like, I don't know. I should do the math more easily in my head, but it's a lot of information. And we've been building various pieces of technology on top of all that data that.
that can point people to right answer, that can understand what's in those images, which is really cool. call that spatial AI, which is the ability for a computer to really understand what's going on in real physical space and give people answers. yeah, that's, I guess that's a bit more of a nutshell, but that's kind of what we were doing. And it really comes out of our, you know, collective experience with really from 3D robotics.
I started building stuff for builders at that time. And honestly, my own scar tissue of running a hardware company. There were so many, we're talking about bedding the company every year, and there were so many moments, I could go into many stories of things that threatened the company's existence because there was something bad happening at the factory. And...
Raj (14:45.4)
Mm-hmm.
Raj (15:00.418)
Mm-hmm.
Jeevan (15:01.449)
I just was like, we'd be on these emails and calls and it was so scary and it was so much stress. And you'd go to the factory and you'd be like, that's the problem? Just put that there. And I was like, God damn it. If I could just teleport into the factory, none of this would be a problem. And everybody's trying to do good job. The factory isn't bad. They're not trying to screw us over. It's just, there's just a failure to communicate, really. And being able to see with your own eyes what's going on was just so.
Raj (15:17.282)
Mm-hmm. Mm-hmm.
Raj (15:27.04)
Mm-hmm.
Jeevan (15:31.305)
so powerful and I just saw that over and over and over again for builders in the construction industry at 3D Robotics. And you know, it wasn't like I was like, I know what to do after 3D Robotics. I kind of wasn't thinking too hard about doing anything. Then Mike and Philip called me and they had their work featured at TED too, this amazing.
Raj (15:32.545)
Mm-hmm. Mm-hmm.
Jeevan (15:51.849)
Amazing talk called birth of a word and anyways any any listeners if you have 15 minutes, you should check it out It's unreal but essentially what those guys expertise was was building technology that consumed absolutely insane amounts of video from regular old cameras And making sense of it and they started working in an idea Around that time Mike had started a company based on that work and sold it to Twitter
Raj (16:00.014)
Mm-hmm.
Jeevan (16:18.985)
We were kind of on the same timeline. He was like getting out of Twitter. I was kind of doing nothing. He started working on this interesting idea with cameras. And as he explained it to me, I was like, you know, could really use that as people, people who build buildings, like, trust me. And that was the seed for OpenSpace.
Raj (16:30.476)
Yeah, so if I can explain to the audience, it's like what I did some studying on this right before this episode, people are mapping out construction sites using a camera on, I guess, do they normally wear on their helmet?
Jeevan (16:48.927)
You can, yeah, that's typically people do is they go for a walk around the project and they put this stick camera on their hard hat. You can, you can hold it if you want to, but it's just very convenient to stick it on there and walk around. This is normal behavior for people that build buildings. Yeah. Yeah.
Raj (16:51.192)
Yeah. Yeah.
Raj (16:59.086)
And it takes 360 video, right? It's a 360 camera. It's a 360 camera. So it maps out this construction site and it helps take like these timestamps, right? Or like records moments in time as the construction project progresses. And then you can always go back in time and see what it looked like at a certain time. And this is a very important in construction, right? But you need to see how things look over time. And also people who are remote need to be able to see how like a job site is looking like at any given time. Is that right?
Jeevan (17:26.069)
Yeah.
Yeah. If you think about how people do this work without the technology that we have, which obviously is how they've done it. Well, it's just extremely time consuming and conflict oriented because you're like, Hey, X and Y isn't done. It's like, okay, is that true or not? don't know. You have to physically walk over there and check. And then if it's covered up by a wall, you have to rip it open. It's terrible. You know? So if you just have really convenient data at your fingertips where you can just go check and see from anywhere, it just makes, makes life easier.
Raj (17:44.983)
Yeah.
Raj (17:48.674)
Mm-hmm. Mm-hmm.
Jeevan (17:57.041)
by the way, mean the idea of photo documentation or building stuff is not new and people have been doing that since cameras were invented, right? All we did is create a better mousetrap that can just not just be a set of random photos that don't know their location but it can totally comprehensive, geo-located, easy to use way of leveraging the fact that a picture is truly worth a thousand words.
Raj (18:04.215)
Mm-hmm.
Raj (18:21.354)
Yeah, sounds good. And like how many I did some like AI searches on like how big you guys are you have quite a few customers now a lot of people using this as well, right?
Jeevan (18:31.165)
Yeah, yeah. The company is like six, seven years old now and we have, yeah, a lot, you know, it's technology used in like over a hundred countries now. And we have like sales teams internationally and just into the U S and you know, thousands upon thousands of companies using it, which of course means lots more users than that.
Raj (18:59.351)
Yeah.
Jeevan (19:00.341)
And these are big, are we target like what we call mid market and enterprise construction companies, owners and trades. So they are, they're not, there are some very big companies that use us and they use us on like thousands of projects at a time. So brands that people have heard of for sure. And also companies that you may not have heard of, but
Raj (19:06.702)
Mm-hmm.
Raj (19:22.723)
Mm-hmm.
Jeevan (19:28.799)
almost for sure built the building that your listeners currently in. You know, if they're not in a home, it's not unlikely that they're building that, one of our customers built. It's cool. It's really cool to see buildings that have used our technology. And there's big ones, know, like Hudson Yards, Los Angeles Ram Stadium, like Marquee things, data centers, but it's a lot of...
Raj (19:33.151)
Yeah.
Jeevan (19:54.263)
A lot of small stuff too. And I think that's been one of the keys to our success as we got, you we knew a bunch of prior to starting open space, right? I should say that at 3D robotics, we built a product for construction companies using drones. So that's how I kind of got into this area. And we basically prototyped the idea and bootstrapped it with some of these builders that weren't building big stuff, just like smaller apartment buildings and even...
We have customers that just do retail fit outs. That was key. It's not just for projects with money to burn on technology. It's just for builders that are stretched in trying to get their jobs done more quickly. That's where we wanted to find product market fit first. And I think that helped us make sure we had a technology and product that was actually scalable and not just for the top 0.5 % of construction projects.
Raj (20:47.086)
Yeah, yeah. And is it deployed via like a special camera or like an app? You have an app, right? Like the people use that. How do you deploy the product? Lesson learned.
Jeevan (20:55.671)
Yeah, yeah. So we don't make any hardware. It's like a pretend hardware company, I guess. We use off the shelf consumer 360 cameras that other companies make. There's a company called Insta360 that makes these. I'm sure everybody's heard of Rico. make 360 cameras. And so our customers just use those. And then our technology consists of a few components. Like there's an app that...
Raj (21:09.869)
Mm-hmm.
Jeevan (21:21.569)
It's pretty straightforward. You you open the app when you're on the project and it's essentially like taking a video. You say, our camera, I hit record, then you walk around and you hit done when you're done and you upload it. And we have a lot of other functionality that allows our users to like do task management, take notes and things like that that are image first. And then of course we web applications. if you're away from the job site and on the laptop, you can go to openspace.ai slash my project or whatever, and then go look around and see what's there.
It was a big, sophisticated kind computer vision backend that powers all this stuff.
Raj (21:54.222)
Yeah, that's really cool. I mean, you guys must be one of the more advanced computer vision companies that are not like self-driving cars, know, like, you know, there's I'm excited to see how computer vision has become like a mainstream thing. Actually, I think I still think computer vision has not peaked like voice. I think everyone is very familiar now with voice. But they're less familiar with computer vision. think like that. Yes, I have is yet to have a chat GPT moment.
But it's gotten really advanced over the last eight years.
Jeevan (22:21.504)
Yeah.
Yeah, well, I think just to opine on technology, when we started the company, we were like, okay, building buildings is very difficult. It's a huge part of the economy. This is a way of making these guys' lives easier, and it's not complicated. It's simple. They're like, yes, this is gonna make my life easier. So we had the seeds of something that we thought could get some traction.
and also we're like, well, you know, if this works, then we are going to, we are going to collect a pretty insane and unique database of, real, real world data, actual real world data, spatial information, visual information. And, that should be good for something question mark, you know, stick that kind of thing in the seed presentation. And, I got to tell you some of the stuff that we're able to do and is happening now.
Raj (23:12.814)
Hmm hmm. Hmm hmm hmm.
Jeevan (23:20.703)
I would not have predicted as possible if you asked me back then, hey, we could do this. I'd be like, nope, doesn't make sense. But there is work getting done that's kind of emerging.
Raj (23:27.79)
Hmm.
Jeevan (23:34.325)
You know, what people think of as AI today, which is like large language models and spatial information and image information. And, you know, we're working on that stuff too. It's pretty, it's pretty incredible. It's like this merging of computer vision and, and, sort of large model AI, like our little phrase that OpenSpaces could call it spatial AI, which is, which is merging spatial computing, you know, which is.
Raj (23:45.55)
Mm-hmm.
Raj (23:55.426)
Mm.
Mm-hmm.
Jeevan (23:59.519)
My simple definition of that is computers being aware of real physical reality, like via cameras and whatnot. And of course, AI, is, you know, my simple, I remember in my, first AI textbook I got, you know, there was a definition of AI. And it was, it was so incredibly vague in a way, but also beautiful because it was vague. was like.
Raj (24:04.174)
Hmm.
Jeevan (24:23.659)
getting artificial systems to do things normally associated with human intelligence. Period. It was not about any particular technique or anything like that. It was just getting computers to do smart stuff.
Raj (24:28.814)
Nice. Yeah.
Yeah.
Yeah, it's beautiful definition, very elegant. just, I mean, the applications of what you're talking about seem like much broader than even construction. Like, you know, you talked about the factory application, but I can think of any physical, you know, basically any kind of physical work, you know, could use this type of technology. I mean, one application that comes to mind is military application. mean,
think every soldier should have like a camera right like like a 360 camera so generals can make better decisions right and stuff like that
Jeevan (25:09.459)
Yeah, I mean, that's one of interesting things about trying to run a disciplined and effective company is focus, but also not shooting too low. And that's a constant struggle, I'm sure. I you've lived that, we all have lived it. And we definitely do have customers at this point that are outside construction, process industries, things like that.
Raj (25:26.796)
Mm-hmm.
Jeevan (25:38.453)
And I'm also like, okay, but you know, depending on how you slice it, you could say construction or real estate is like the actual biggest part of the economy. There's plenty of problems to solve and plenty of market to go get. So keeping our sales team and go to market focused on the one area is probably best for us. But you're totally right. know, like basically anyone's, anyone whose job depends on managing a real physical reality.
Raj (25:51.342)
Mm.
Jeevan (26:07.959)
like making sure it's in state X or needs to go from state X to state Y, it would benefit, it benefits you to like have this information at your fingertips so you can see how the world is changing without physically walking around and like writing a report or whatever. You know, a lot of people who have this job, if they're building a superintendent or construction, and this is kind of a nerdy way of putting it, but hopefully listeners appreciate it. You think about like writing a report about even a home inspection.
That is an extremely lossy warped information compression thing. Like you are taking the data that is the physical space and you're turning into text and much is lost when you do that. And honestly, it creates a lot of problems because it's not that trusted.
Because you don't really know someone says this or that you don't know what they're talking about half the time and then you may not trust it. And so what do you do? Okay. I got to go get myself all this kind of stuff. And it just takes also a lot of time to create those types of reports. And so if you can just disintermediate that it's like, don't write a report. don't need to write a report anymore. Just go log an open space and look and see for yourself. And then you can write tasks and stuff within it, but you're just, you're just.
like saving so much time. It's just a nice, it's just a nice execution layer. And obviously there's always gonna be reports and there's gonna be drawings and documents of record and 3D models and all that stuff. But if you can have this visual project execution layer, like we believe it just is gonna simplify people's lives. It does, and our customers tell us all that all the time in a way that isn't, it's not evangelical. It's not like, hey, there's a brand new world and you have to see how OpenSpace is doing and then.
Raj (27:45.582)
Mm-hmm.
Jeevan (27:52.703)
Understand that people get it within five seconds
Raj (27:55.278)
Yeah, it seems super valuable. mean, 3D cameras had a moment with consumers at some point too. I think there was like a few years ago, right? But I don't think people really use it that much in consumer land anymore. do you think there's any consumer applications for this type of thing as well? There used to be this idea of drones with cameras, And then they were taking 3D. But that seems to have gone away. Like no one's using drones.
Jeevan (28:06.081)
Yeah.
Raj (28:25.568)
as much, right, to take photos here.
Jeevan (28:25.739)
Well, it's funny, some stuff becomes invisible because you don't think about it anymore. You don't read about it, but it's very much just because it's become part of the fabric, but people just do routinely. you know, so I'll take 360 cameras. Yeah, we use them and they have, you know, these very valuable applications for enterprises and people still use them if they're going surfing or whatever, because it's a cool way to like capture what you're doing and then maybe able to edit it down to what you want.
Raj (28:50.754)
Mm-hmm.
Raj (28:55.63)
Mm-hmm.
Jeevan (28:55.991)
And then drones, mean, you know, 3D robotics was a drone company and I guess it still is. It was eventually acquired by Esri, which is a big GIS or geo information systems company. you know, like the use of drones is it's not, I wouldn't say it's routine yet. in certain industries.
you know, like mining or construction, but it is much more routine than it was five or 10 years ago. Like these just happens. You look at Skydio, which is a great company, and other MIT guys, and you know, that continues to grow. And a lot of it is outside the boundaries of consumer applications.
Raj (29:23.968)
Mm-hmm.
Raj (29:28.174)
Mm.
Mm-hmm
Raj (29:35.886)
Mm-hmm.
Raj (29:41.249)
Mm-hmm.
Jeevan (29:41.429)
which I think is where the real value can be. But there's still no people that are just flying drones around for fun too. It's just not like, it's one of those things where technology kind of matures and then you just sort of forget about it. Like in the early days for drones, just making it easy to fly was hard. Making it so it doesn't crash into stuff was hard. Making it so the camera was steady and the image quality looked good was hard. But all those problems kind of got solved.
Raj (29:51.95)
Mm-hmm.
Raj (30:00.158)
Mm-hmm. Mm.
Raj (30:06.67)
Mm-hmm.
Jeevan (30:09.321)
And now they just kind of work fairly well and that's out there. You know, they're just out there. If you want to buy one, you can get it. It's not a, it's not that expensive or that hard anymore.
Raj (30:12.493)
Mm-hmm.
Nah.
Raj (30:20.366)
It also strikes me that there's a similarity mentioned like another MIT company like samsara, know, do you follow those guys? I mean like they're instrumenting a physical industry, right?
Jeevan (30:27.403)
Yeah. Yeah. Yeah. I mean, I Samsara, like there's lot of spiritual alignment between what Samsara has done. Obviously they're bigger than us and they've they're public now and all that kind of stuff. But it's a similar idea. You know, I was talking to Amir Varani. I don't know if you met him, but he was one of the founders of Dropcam. If you remember Dropcam, you know, it was acquired by Google and became the Nest camera system. And
Raj (30:49.54)
yeah, I Dropcam, yeah yeah.
Mm-hmm.
Jeevan (30:54.737)
I don't know if he said this or something, I remember in conversation, he was kind of, he said something like, you know, if you put a camera where there wasn't a camera before, it's probably a good business in there.
Raj (31:04.686)
Yeah. Yeah. Yeah.
Jeevan (31:05.911)
And I think that's kind of true, you know, because images are good. People get it. And, um, I think Samsara has done that. It's like, okay, for fleets. So if we really don't know Samsara, like I think their main thing that they got going on was just putting cameras and trucks and fleets. So you can see what's going on. You know, um, it's a good idea. It's a good idea. You know, I'll say one other thing about this conversation with Amir when Dropcam was, and they were growing fast off, so they did well. And I remember looking at.
Raj (31:14.382)
Mm-hmm.
Raj (31:21.686)
Mm-hmm. Yeah.
Jeevan (31:35.381)
the DropCam website and it showed like, and so DropCam was like, you if you look at Nest camera, all these sort of consumer cameras that was, they kind of figured that category out first, what was to do it really, like a web connected camera for consumers. Anyways, you went to their website and it was just like, there's a camera and it says like, you know, I forget, I'm gonna butcher this with something like simple home camera for whatever.
Raj (31:46.318)
Mm-hmm. Mm-hmm.
Jeevan (32:01.085)
And I was like, hey, you know what's kind of weird about your website? Like you don't, you're breaking every sort of rule of marketing. You don't explain what it's for, like why people would want to use it. What's the value proposition? He's like, yeah, yeah, it's true. But people kind of just know what cameras are for, you know? And I was like, it's true. You know, like, I think people kind of know, what the value of getting an image image did it is. So if you can make it extremely convenient to.
Raj (32:15.694)
Thank
Jeevan (32:31.113)
generate imagery in a place where people would like it, that it was hard to do before, you can really start getting momentum for that. And obviously they were successful with that too, and Sam Sarris as well.
Raj (32:43.094)
Yeah, no, that's such a great statement, you know, if you put through a camera where there wasn't one before it can probably build a great business That has actually been huh? Yeah
Jeevan (32:49.737)
Drones are not, by the way. I mean, obviously they're shipping drones, but a lot of drone application is putting a camera in the sky. I mean, that's what it is. And that's useful.
Raj (32:55.66)
Yeah Yeah, that's right I've had this thesis for a long time probably for 15 years that cameras will become you know ubiquitous like basically like in every single corner of our world and I always thought there would be a lot more public as well like there'd be cameras on every street corner You know, and now there's a company called flock safety who's like putting license plate recognition cameras everywhere It just seemed like a no-brainer to me. I mean like of course, I mean my where I live in Los Altos has them now too is because
Jeevan (33:05.493)
Yeah.
Jeevan (33:17.291)
Yeah. Yeah.
Raj (33:24.394)
And people want him for security, They want to know if someone came by their door to pick up their Amazon boxes. They want to who that person is, right?
Jeevan (33:33.015)
Yeah, I agree with that. think it's happening. it's funny. Like, I think one of the reasons I got involved with Lux Capital was, you know, at the time, you know, was kind of not doing anything. It was great. My kids were littler, taking the zoo every day. you know, was deciding what I wanted to do next. Maybe start another company, maybe not, whatever. And I had free time to like have my own.
Raj (33:55.886)
.
Jeevan (33:58.615)
personal, stupid blog as one does, you know, and have various random musings. And one of them that really inspired OpenSpace, you know, I...
Raj (34:01.894)
Mm-hmm.
Jeevan (34:13.271)
You know, I called it camera ubiquity. And my basic thesis was, if you think about the kind of Moore's law thing, lot of people express that in terms of like number of CPUs per human at a period of time. And obviously in the fifties, it was like one CPU for like a hundred million people or something. Right. And then gradually that number, that ratio started to change. And now of course today it's like.
You know, how many, how many processors are devoted to each individual's life? It's some crazy number. mean, who knows? Like right now I have many just on my desk saying nothing out of the cloud. Okay. So if you sort of take that as like a ratio that changes, I posited that the same thing is true for cameras and cameras just a little further behind. So now some cameras were first invented, there are very few cameras and they were very precious things.
Raj (34:51.822)
Mm-hmm.
Jeevan (35:07.605)
And then you got to the point and shoot cameras where you might have an individual may have one camera, right? But now it's many, like your phone has many, you know, cut computer does, then they're kind of getting everywhere. And my basic argument was that heralds a new service area for technology and spatial computing to go back to that, where you can start building products that are aware of real physical reality.
Raj (35:13.473)
Mm-hmm.
Raj (35:29.006)
Hmm.
Jeevan (35:34.647)
You don't need your customer to buy this expensive thing that's hard to hook up and annoying. Or video conferencing is another good example. Remember video conferencing used to require a bunch of gear and you'd only see it in big company boardrooms and now it's built in a re-computer.
Raj (35:49.07)
Mm-hmm. Mm-hmm.
Jeevan (35:54.623)
I believe that and I think that's part of what made OpenSpace even possible is connecting a camera to the cloud as possible. Anyways, heard up a little thing about that and I remember talking to Lux guys at some CES party and they were like, hey, you know, should be in EIR here because this is an interesting idea. And I was like, okay, sure, why not? I'm not doing anything else. And that was like one of the things that OpenSpace idea got sucked back in. Yeah.
Raj (36:02.178)
Mm-hmm.
Raj (36:11.646)
Mm. Mm-hmm.
You got sucked back in. Thanks to your blog. Yeah. No, it's great. I wonder where we are on this curve. I still feel we're early. I still feel like there's a lot more cameras to come in our lives. In general, sensors. Yeah.
Jeevan (36:32.149)
Yeah. I think we are early. I think you're right. think, like, it's still early because, yes, there's a bunch of cameras for you and for me, like, in our pockets and whatnot, but...
I mean, I'm sure there's a lot of creative entrepreneurs out there that figure stuff out. There's the cameras that will get inserted into infrastructure. And of course, I think that's that raises lots of privacy things too. And so there's there's like cultural things to be solved or not solved. People may not be comfortable with that. Yeah, but you know, we'll see how it goes. I mean, even in San Francisco where I live, like, I think there was some resistance for SFPD using drones.
Raj (37:04.856)
Yeah. Privacy is dead, you know, they're not going to be.
Jeevan (37:19.977)
I think that's kind of gone. I mean, I'm sure some people have, there's various opinions, but if you look at the actual behavior,
This a pity does use drones and Lucy I talked to her like that's good because smashing grabs are super terrible and this is a way to catch guys and and I think they've been very good at marketing it, know, putting these videos on Twitter of like a drone view looking at like a bad guy following it around showing the police arrest the guy. So there's like a there's a public perception thing being worked on.
Raj (37:29.3)
Mm-hmm.
Raj (37:44.001)
You
Jeevan (37:53.215)
So I think that there's the infrastructure, there's cameras getting into robotic systems, of course. And then I think about, the way I think about it is cameras used to be, it used to be the endpoint of camera data. was a human looking at images or video. But I think we're shifting where.
Raj (38:00.558)
Mm-hmm.
Jeevan (38:20.971)
Cameras are consuming information where the human will never even look at the images or video. It's just too much, but they'll look at a summarized version or some answer that they want that's compressed in some way, maybe down to text or a montage or something. And that's like, that relies on...
Raj (38:32.184)
Yeah. Montage maybe. Yeah.
Jeevan (38:43.145)
on the continuing improvement of cameras and it relies on advances in computer vision and artificial intelligence to reliably translate a massive amount of images into an answer to a question. But at OpenSpace, that's kind of one thing that we do and we work on. So if you are, we know we had a little demo of this at our user conference recently.
But it was a simple idea. It's like you are building a building and you want to know where your materials are. Like is the drywall there for the drywall folks to install it? Could you just ask OpenSpace where the stacks of drywall? And it just knows. And you know, we're getting to that point. So that's pretty exciting. Or you want to know if people are on schedule, you know, and could you take all the images and just say like you're on schedule or not by just analyzing everything that's in there.
Raj (39:34.466)
Yeah.
Jeevan (39:36.087)
And the truth is that's a very sophisticated, difficult problem because even skilled humans have a hard time making that judgment. But, you know, even combining human and artificial intelligence to get to the answer is very interesting to me. So I agree. I think we're early. Like cameras can get sophisticated to get connected and our ability to process video data and spatial information is, is we're just...
Raj (39:37.038)
Super cool.
Raj (39:51.608)
Totally.
Jeevan (40:02.389)
toying with it. It's not like in five, 10 years, it's going to be ridiculous with what we're able to build.
Raj (40:08.334)
Yeah, very exciting. Very exciting. I don't know if I told you that at Presto, we also had a thesis of computer vision. So we used to think that there would be cameras everywhere inside and outside a restaurant and you'd be able to map every single operational touch point. like every single like every time the waiter came to your table, you'd be able to know, you know, you'd be able to measure that. So you'd be able to measure people keeping to spec, you know, on like on waiter touch points. How long did it take them to clean the table after you left? You know.
Jeevan (40:18.505)
Yeah, yeah.
Jeevan (40:22.175)
Mm-hmm.
Jeevan (40:27.062)
Yeah.
Yeah.
Jeevan (40:33.537)
Right.
Raj (40:34.99)
You can even you know monitor like people cooking things correctly in the kitchen like every single like you know because because restaurants are like so if it's their factories and their retail environments at the same time and And then the drive-thru is of course what we were doing. We were we were Recognizing cars so we were able to recognize repeat visitors and then you can like tailor you can actually like also like You tailor the sales basically to the car. So if it's like a more expensive car
Jeevan (40:44.341)
Yeah. Yeah.
Jeevan (40:53.397)
Yeah.
Raj (41:03.586)
You could actually tell, give them a more expensive upsell versus like a cheaper car. Like there's a lot of things you could do. And I still, I mean, that application I think is very powerful, but the customers and restaurants were a bit slow to adopt, to be honest, right? Like it wasn't the most sophisticated.
Jeevan (41:07.627)
Yeah.
Jeevan (41:19.211)
Yeah, totally. It's, and it's about what are the actual pain they're experiencing. I'll tell you, that's interesting. You know, I worked at McDonald's in high school, the drive-through that's, you know, shout out to the people that work at drive-throughs. That is such a difficult job. Like the cognitive load is insane. and so just relieving the pain of a drive-through worker would be great. Now, obviously drive-through people working the drive-through are making the corporate buying decisions at.
Raj (41:30.35)
you
Jeevan (41:44.597)
McDonald's or whatever, but you know, I feel for your drive-through people. I've been there. But yeah, I think it's all about relieving these pain points too. You know, one interesting thing at OpenSpace is like the, the pain of just not knowing what the hell is going on is very real. People feel it. If you can, if you can address it, then that's good. But then there's all these other things you can add to it and just finding the time.
Raj (41:49.621)
What what?
Jeevan (42:08.073)
in your customers mind and in your own company's journey to insert these things is not easy. But one thing that as we get higher, you know, get more more adoption, you start selling to and communicating with the C-suite, which is cool. And what their pain points are related to, but they're kind of can be distinct from the front lines. And the same way that.
Like a waiter may not care about the operational efficiency, maybe if you own a franchise of like 80 chilies or whatever, you would. One example of that is us as insurance. And this is cool too, I would say another way that Canvas Computer Vision might get more deployed is we have this thing now where we can say to our customers, if you use OpenSpace,
He will literally save money on your insurance policies. And there's lots of complex insurance policies for construction companies. And the reason is, is there's an insurance company that just studied the use of open space versus not open space building projects. And they found that the claims frequency is way less on open space using.
And if there's a, there could be selection bias in there that people use our technology are inherently good builders. So they also measured a given builder before and after using the technology and watch the glimpse frequency go down a lot. And so this insurance company called Shepherd, very interesting company said, okay, we can underwrite that.
we have competitive advantage because we can make insurance cheaper and that's like a no-brainer for CFOs of construction companies. The reason I bring that up is you look at these applications of computer vision technology that have all these potential benefits.
Jeevan (43:59.531)
and then what actually lands with the customer and just trying to drive that into something that is tangible for them is really cool. So, you the idea that like a computer vision or AI or software company could literally make it less expensive to build a building is kind of crazy, right? I mean, it doesn't make sense, but we've actually done it. And so I feel like those are some of the unlocks across industries when you can make those moves that can really create a lot of change.
Raj (44:22.466)
Yeah.
Jeevan (44:27.891)
and adoption of technologies for, because it just makes sense to everybody. Everybody in that company can say this is a good thing we should use it.
Raj (44:36.01)
Yeah, no, it makes a ton of sense and i'm excited for You know the work you're doing in this space and I think there's there's a lot of growth Potential for what you're the technology you're building. So it sounds like a lot people could probably benefit from Some of the spatial ai stuff. I think there's work going on in multiple industries in this On the front robotics is definitely one and then you know self-driving cars You know, but I think in some other industries is very It's very nascent. you know and like I invested in a company
Jeevan (44:54.092)
Yeah.
Jeevan (45:02.496)
yeah.
Raj (45:05.614)
called Protex, did, or they're still doing, I mean, they're doing a health and safety analysis for warehouses. So like they put cameras in warehouses and they're basically trying to like capture potential dangerous situations. Again, maybe it could be good for insurance, but also just for the management there. People need.
Jeevan (45:12.033)
Yep.
Jeevan (45:17.655)
Mm-hmm.
Raj (45:23.368)
Becoming a compliance thing right like they need to feel like they need this But in general like you can imagine almost every work site or every job site will every place where people work It's gonna be full of cameras because there shouldn't be privacy in those places. Those are places are for work, right?
Jeevan (45:25.226)
Yeah.
Jeevan (45:33.601)
Yeah.
Jeevan (45:37.335)
Yeah, well, that's a really interesting thing too is like when they were getting open space off the ground is this privacy concern of people like, no, I don't want these cameras. But I think details matter. Like any business details matter. It's an idea that makes sense. And when that doesn't can come down to the details. And for us, I think there were kind of two, well, you just ask these guys like, is this, would you like it? And if they're like, no, then don't start the company. Right? So we definitely did that. But, um,
Raj (46:03.735)
Hehehe.
Jeevan (46:06.645)
And I what it boiled down to for us is with open space, like as we were talking about earlier, it's not constantly on. It's just someone walks around with a camera, they pop it on their hard hat, they do their walk, they take it off. And so people feel like they have control. And that's not just a privacy control thing. The truth is, like sticking lots of cameras everywhere on a construction site is kind of not practical because do you have power or not? The thing is getting built. Well, anything that's...
Raj (46:17.038)
Mm.
Raj (46:34.029)
Mm-hmm.
Jeevan (46:36.095)
that's at a job site is in danger of getting destroyed. So there's a practical thing to it. And then also what you said is totally right. It's a work area. There are already expectations that if you're the general contractor and you wanted to take photos of what's going on, the trades are not going to be like, no, you can't, you're forbidden from that. And vice versa, if trades want to document their work.
Raj (46:39.502)
Yeah, of course, of course, yeah.
Yeah.
Jeevan (47:05.611)
that makes sense if they have the right to do that. So I think a lot of those questions kind of go away, both due to the fact that it's a work area and not like a consumer public area, and also the details of actually how you're collecting these images. Is it gonna make people get their hackles up or not?
Raj (47:25.006)
Yeah, yeah. Another thing that I think about is the meta ray-band glasses, their key feature is a camera in the glasses, So cameras are getting smaller and being inserted in all sorts of wearables as well, I guess, at this point.
Jeevan (47:32.629)
Yeah. Yeah.
Jeevan (47:41.823)
Yeah. Yeah. The meta-rate balancing is really interesting. mean, have you seen just people wearing those around yet? mean, I've like among tech weirdos, like I know, like such as myself and co-founders and whatnot, you know, like we're definitely experimenting with them. I I'm very curious to see if that gets wider adoption or not. You know, it's, it's like the details matter. it like, cause you know, snaps experiment with this and like, the camera.
visible or not and things like that. I don't know how to the future on that, but it's very interesting.
Raj (48:21.248)
Mm-hmm. Yeah, I don't know. haven't seen too many people walking around with a camera on because the light does come on so I that's how you know But I've heard from other people that they use it to take photos of their kids or videos of their kids when playing with their kids Right, which I can see because kids at certain age get very uncomfortable or they get very conscious self-conscious when you point a phone at them
Jeevan (48:27.285)
Yeah.
Jeevan (48:34.325)
Yeah.
Jeevan (48:41.387)
Right. Yeah.
Raj (48:42.732)
So actually to have this, would solve a problem for me for sure. ironic, my prescription is too strong. They don't want you to support my subscription, my prescription. So yeah.
Jeevan (48:46.23)
Right.
Jeevan (48:49.815)
Yeah, you need to get LASIK. You need to have a LASIK plus Meta-ray bands bundle deal so you can...
Raj (48:55.598)
Exactly. Yeah, yeah, yeah. But I know it's an interesting, you know, again application of cameras and you got to imagine there'll be more sort of wearables and people do may want to record video of everything they do in their daily lives, you know, but of course you'd yeah.
Jeevan (49:10.539)
Well, that's like that old, you people talk about in the fifties. What was it? I feel like I'm a bad MIT student. It's Gordon Bell's project. Cool name. The idea of wearing a camera at all times and just recording everything. I mean, there's definitely like, what is it? Black mirror versions of that. And then there's like ones that are maybe less dark.
Raj (49:20.44)
Yeah.
Raj (49:26.935)
Okay.
Raj (49:34.156)
Yes, yeah That was the contact. Yeah, the contacts that could record video and you can go pay back like old stuff. Yeah. Yeah
Jeevan (49:39.583)
Yeah, yeah. But it's very interesting. mean, like, that's the cool sci-fi stuff is like if that existed, what world would we live in and what are the what would be the benefits and what would be like the sort of unintended consequences for interesting to think about.
Raj (49:44.364)
Hmm.
Raj (49:54.806)
Yep. Well, on that note, I think we should wrap. This has been a fascinating conversation. Jeevan, really appreciate you coming on and chatting about this. It's, again, I think with one of the underhyped areas of AI, surprisingly, you know, and lots more to come.
Jeevan (50:11.093)
Yeah, well, it's absolutely my pleasure and it's good to reconnect with you and get a podcast done at the same time. So thanks for having me. Yeah, yeah.
Raj (50:17.458)
Yeah, exactly. It's an excuse to catch up for sure. Yeah. Thanks, man. 
Share this post