%20copy.png)
Zach Rattner: The basic reason that we exist is we can compress down these tedious and complicated in home inspections that can take hours and cost billions, and we've shrunk it down into a process that takes just, ten, fifteen minutes.Song: Built this week, breaking it down. Built this week, we show you how. A fresh idea, a clever tweak you locked in. You built this week.Sam Nadler: Hey, everyone, and welcome to Built This Week, the podcast where we share what we're building, how we're building it, and what it means for the world of AI and startups. I'm Sam Nadler, cofounder here at Rise Labs. And each and every week, I'm joined by my friend, business partner, and cohost, Jordan Metzner. I'll introduce Jordan. You may notice that this episode is a little bit different.We had a little bit of technical difficulties, had to record the intro. And with that, Jordan, how are you doing this week?Jordan Metzner: Yo, Sam. Happy to be back. This is a great episode since we've already recorded it. We have an awesome guest who's, you know, really done a great job using AI to implement some new technology in in an old, antiquated industry. So super excited about our guest this week.Obviously, it's an incredible exciting week as Fable has come back to the developers. And, yeah, it certainly kept me up late at night. So, anyway, really looking forward to getting into this week's episode, and let's kick it off.Sam Nadler: Yeah. And I'd love to introduce our guests. But before I do, don't forget to like and subscribe. We're a few subscribers away from the 30,000 mark on YouTube. And with that, I'd like to introduce Zach, the CTO of Yumbo dot ai, and thank you for joining.Zach Rattner: Sure, Sam. Thanks for having me. Happy to be here. My name's Zach. I'm the chief technology officer and cofounder of Yembo.And Yembo builds computer vision technology that deeply understands the interior of homes. We sell to property insurance companies and moving companies. And the basic reason that we exist is we can compress down these tedious and complicated in home inspections that can take hours and cost billions across the whole industry. And we've shrunk it down into a process that takes just, like, ten, fifteen minutes. And we use computer vision to deeply understand the interior of the home.If you're moving, we can make an inventory list. If you're doing a reconstruction estimate with your insurance claim, we build a three d model of the space, floor plans, and we basically are bringing AI in to simplify and streamline this process and help companies run tighter shops overall.Jordan Metzner: That's great. And, what type of companies are using the tool?Zach Rattner: So we have, companies all across the world in our moving product. We're deployed in about 35 different countries across the world doing thousands of inspections every single day. Concentration, obviously, in North America, but we do have good presence in Europe and Asia, Oceania as well. The property insurance business, as you may imagine, it's more heavily regulated, so we're US only for right now. Although, are actively scoping out pension opportunities all across the globe there.Jordan Metzner: That's awesome.Sam Nadler: With that, Zach, let's go into the demo. Why don't you share a share a quick demo of how Yoimba works?Zach Rattner: Sure. So I have a demo pulled up. Let me quickly explain, then I'll I'll screen share so you can see what you're looking at. So imagine you are moving your house from New York to London. Traditionally, that would involve calling a company, scheduling some time.Someone will come by, ring your doorbell, and then note down all the items that need to be moved. That's important for logistical reasons, like what size truck needs to be brought, how many cartons to pack up the ducks that you see behind me, things like that. And then also for international ones, you gotta deal with customs. For domestic moves, you have to deal with DOT regulations. And this process can take sixty to ninety minutes.And what we've done is we built technology where rather than having an in person estimate happen, you can have the link texted to you, and on your phone, you can go through just in your web browser, scan quick videos of what's there, and the AI will identify what's in the room, make that inventory list, and not just give you a handwritten list of items, but the actual pictures of what's there. And this is important for, this is important for preparedness on the day of the move, making sure the crew brings the right equipment so you don't have exceptions on the day of the move. It's helpful as a checklist. You can go through and make sure you've loaded everything and then unloaded everything. And it's also important if you do have a loss or a damage claim to go back and see what was agreed upon, being able to compare the item that's been damaged to what it was before the movie even happened.So let me go ahead and share my screen, and I'll show you what the output looks like. So we call this the visual inventory. This is the report that our AI generates. We give you a summary of all the different packing materials that the AI automatically calculated from looking at all the items in the room, the items that are moving, not moving, and the part that is, like, the main piece the AI generates is these film strips of images. So you could see here it's not just saying, hey.There's a bed and a lamp in the bedroom. You can actually see color coded and numbered all the different items that need to be moved. What's moving? What are the packing materials? And when something's marked as not moving, it's date and time stamped in here too.So when we first launched this, we had a customer that, was moving from Washington DC to London. And when you do international moves, it could be very complicated. There's companies on both sides of the of the equation that get involved, and there was a mover at the destination that was picking up the shipment from the port, and their job was to unpack it and bring it into the client's new home. And they saw this visual inventory by Yenbo saying we got this funny email that just said, I got one of your reports and who on earth has the time? And the funny part was this is, like, 2020 or so, like, before AI is as it is today, and they thought that somebody actually went through and took pictures and got the colored numbers on it and spent probably, I don't know, five, six, seven, eight hours, putting this report together by hand.And then we told them that that actually happens on every single move. We use this new technology. It's called computer vision. AI did it. It wasn't a human.And I feel like that value proposition of AI, the promise of you can do something that is better than the way work was done before is really empowering. I think the obvious thing to do when people hear about AI is to automate and replace and just do the same old things with some percentage faster. But I think maybe the more interesting and compelling thing to look at is how can you use the fact that these things have become easier and faster and then go do something that is higher quality or to go do something that would have maybe been able to have been done before but is really cost prohibitive or labor intensive. And I think this is one such example, but the whole sector has, I think, many more on what you can do when you bake AI into the workflow. Kinda lets you rethink exactly how everything works.Jordan Metzner: I think, you know, what makes Yamba so interesting is that first, you're talking about an industry that's very antiquated, very, you know, kind of lack of technology in general. Right? It's very handwritten industry, a lot of paper, things like that. And then the second piece is that your software integrates with not even the person that uses it, it integrates with kind of like the homeowner, which is kind of a flip on how things used to run. And then the outcome is that the actual like mover itself gets the advantage of AI without them themselves even never needing to even know that AI exists or how it works.And I think that's like actually like a pretty magical like formula here. Because if you were to say like, hey, I'm gonna give AI to to movers, you know, you would probably fail pretty quickly because like they're not, you know, technology first. They don't use, you know, the technology in their workflows. You know, they don't have a computer with them, and they, like, do this, like, you know, manifest or inspection, etcetera. And so, like, it's so interesting to flip the model on its head where the mover itself doesn't actually change his behavior, except that now he has, like, even better paper document.Just, like, takes his job to the higher level without ever, to your point, like, who has the time to put a report of this quality together by hand. Right? So awesome. Awesome demo.Zach Rattner: Yeah. And I think it's really, frankly, a bigger challenge than probably we talked about at that point where a lot of people who use AI use it every day, and they'll spend inordinates amount of time using it. I don't know about you, but I I got Claude open from the moment I get to my desk to the moment I leave, and then sometimes after I leave, I open it on my phone. But in these workflows, there's a job to be done. Nobody wants to deploy AI just because it's AI.Like, they have quotas they need to reach and work to be done. And in our cases, the people who are actually using the AI have never done it before. And if you're the homeowner, you may never do it again or maybe ten years before you use it again. So building the technology in a way that is simple and intuitive enough that you can just pick it up and use it is much easier said than done. I think probably 30% of my company's patents are on what do you do during the camera capture process to make sure that the AI sees what it needs to see, but without making somebody, like, watch a two hour tutorial video or read a manual beforehand.I think it's sort of sort of overlooked in a lot of these cases that when you deploy AI into the physical world, a lot of times there's not a mature infrastructure around people using it, their awareness of what the underlying technology does, even, like, clear limits on what the technology can and can't do. And it makes opportunities for folks to figure it out, but I'd say it's anything but obvious when you start what that right answer is gonna look like.Jordan Metzner: Well, that's great. No. I mean, you know, people just wanna get their stuff moved, and most people to your point will kinda like know how to use an iPhone or an Android phone, but they have no idea how to take a home inventory prepared for a professional international move. And if you can let, you know, someone do that in ten minutes on their phone, it just saves a lot of time and money and it seems like probably headache down the line as well. So, yeah.Again, I love the demo. I think it was awesome.Sam Nadler: What used to take six months now takes six weeks if you have the right king. Rise Labs pairs AI native engineers with your business to move at startup speed. Build faster, launch sooner, learn in production. This is Rise Labs.Jordan Metzner: Okay. Sam, why don't we, why don't we jump into the news?Sam Nadler: So as you guys recall, Fable five was pulled from the markets. I think as of last night or early this morning, it has been authorized for use, but as as I understand, not for coding. I know, Jordan, you're a you in the thirty six hours, you were able to use Fable five. You you loved it. Zach, I don't know if you've had much exposure, but now that it's back, any any high level thoughts?And, you know, are we going to continue to see the government get involved as as new models or, powerful models come out?Jordan Metzner: I'll go first on this one. Yeah. I mean, I used table five when it came out, and it was amazing. And then they took it away, sadly. And I would say that the current models are not nearly as good at 4.8 even with the, like, ultra code as as well as 5.5.And obviously, you know, we've seen 5.6 be announced from OpenAI to a limited audience. I think they said, like, 20 companies and, obviously, he's also working with the government. So I think, you know, we're probably at the point in which these powerful models will probably continue to go through this type of self assessed process. I think it looks like even without regulations in place, these frontier labs don't want their models taken down by the government. And in order in order to avoid that, then they need to be, you know, overly friendly with the government.So I think it seems like they're creating their own framework at the same time that, you know, the framework doesn't really exist. But as a developer and programmer, I wish I got Fable five back. But I think the news is the little misnomer because what it said was Fable five is coming back, but not for coding purposes. So at least that's what I read, that it is gonna roll back to 4.8 in that case. And like, you know, he was mentioned a 5.6 as well.So, yeah. From my perspective, want Fable five back. But if I can only use it for queries, then, you know, it's only so good. But if I can use it for coding, then I'll be pretty happy because it did did do an amazing job. I don't know, Zach.How about you?Zach Rattner: Yeah. I mean, in my view, I think the AI industry is sort of reaching adolescence. If you look at other markets like the automotive industry, it would sort of be unheard of to launch a new car, but then not have safety testing done or to launch a new drug in the pharmaceutical world, but then not exactly know the implications. And, I mean, I do sympathize where if you are in a brutally competitive market, everything's moving at light speed. US companies compete against each other, but I think also we're competing against China.It's a lot of pressure to make a lot of progress as fast as possible. So I'm not particularly surprised that as these models get better and better and better, they become more and more capable. And at some point, some regulators says, hold on a minute. Like, have we thought about what's really going on here? So I'm not particularly surprised by this.And, honestly, I'm, encouraged by OpenAI's decision to not do, like, a launch yank launch again. Like, at some point, it had to happen, but I wouldn't mind seeing some clarity around guidance on these sorts of things. I think it's important. They're powerful. They're becoming a part of everyday life where it is disruptive if you yank it.Where two, three years ago, if you yanked it, a lot of people lost an interesting tool, or a toy. But now if you yank it, people are losing their ability to get real work done. I'm glad that we didn't have dependencies built up yet on Fable, because I think if these sorts of things can happen, then you'd have a lot to rework and fall back on. But I do think that it's sort of that next stage as we enter into AI adulthood, if you will, that we sort of think about these things more of like a public utility unless there was, like, a novel interesting toy. So it's unfortunate to slow things down, but also I wouldn't wanna live in a world where no cars were safely tested before they hit the roads.Jordan Metzner: Let me just play a little devil's advocate there because I I don't think I do totally disagree with you, but, you know, I think, you know, what part of this is, you know, just these companies trying to move fast to grow revenue versus kind of really doing a lot of deep KYC and understanding who the customers are and using their tools. Because, you know, if you think of, you know, I think the equivalency of looking at, like, know, the top tier clouds. Right? You know, you can't easily go on AWS to launch a server or scale of servers as a no name user that just created an account in the first day, you know. And even just to get a bigger server, you need the permissions and you need acceptance.You know, this is a whole series of controls put in place in order for you to do something that, like, could be somewhat dangerous. And, you know, obviously, these are startups that are VC backed and are, like, growth at all costs. And I think I think the news said something like they found, like, 25,000 Chinese accounts, like, hacking you know, basically, like, hacking onto the model. That doesn't sound like they put, like, great KYC in place and, like, IP checks and really, like, know who their customer was. So, you know, how much of this was just, like, maybe these laboratories being a little flagrant with their behavior, and then this happened with a powerful model.And, you know, have they put, you know, more controls in place earlier on, maybe wouldn't have seen this type of behavior. But, you know, I don't know. I mean, like, self regulation might be the best. You know, I don't know if regulated industry, to your point, like, is AI like, you know, like pharmaceuticals, a car, you know, or is it more, you know, like a, you know, like a coding language or a database, you know, technology or something like that that, you know, shouldn't be regulated. I don't know.For me, it seems too early to tell, but I I hear your point.Sam Nadler: I think you both had really interesting points. It feels like we're getting into the adolescence era. It feels like these models are becoming more and more powerful, and, you know, having a rollout framework makes a lot of sense. It doesn't seem like it's really there yet. But, yeah, I think, you know, as as as these models become more and more powerful, just something something urgent or something defined needs to be created.Jordan Metzner: Awesome. Thanks, doc. How can people get ahold of you or find out more about the company?Zach Rattner: Sure. So the company's name is Yembo, y e m b o. I intentionally made up a word that doesn't mean anything. So if you Google it, we will be number one. You're welcome.That was a good idea I made ten years ago. Not every idea I've made has been good, but that one I'm proud of. I'm also pretty active on LinkedIn. So if you look up Zach Ratner, r a t t n e r, that's the only real social network I'm active on, so you can feel free to connect with me via note. Always happy to chat about AI.Jordan Metzner: Thanks for coming on. Thanks, Tim.Sam Nadler: Zach, thank you. It was a pleasure. Great episode.
