Pod Insights This Week in Startups Decart makes AI faster, Lume teaches lamps to fold laundry | E2166 About this episode Today's show:
We're back with two insightful new TWiST founder interviews.
First up: Dean Leitersdorf of Decart tells us about squeezing maximum productivity out of your GPUs. But it's not all talk: he also shows us the incredible open world model that can magically transform live footage.
THEN! Jason and Alex chat with Syncere AI founder Aaron Tan about Lume, his robotic lamp device that went viral for folding laundry. Hear why Aaron thinks the future of robotics is not necessarily humanoid, and all about his future plans for the Lume arms.
Timestamps:
(0:00) Alex introduces our TwiST 500 interviews
(02:27) TWiST 500 Interview #1: Dean Leitersdorf, CEO and Founder of Decart
(04:37) Setting out to build a "kilo-corn," a trillion dollar company
(06:27) How making AI much faster opens up a world of new opportunities
(09:48) OpenPhone - Streamline and scale your customer communications with OpenPhone. Get 20% off your first 6 months at https://www.openphone.com/twist
(11:00) Why Dean thinks chatbots are the interface of the future
(17:24) How Decart builds and trains its models more efficiently (and for less money)
(20:01) Netsuite - Download the ebook CFO's Guide to AI and Machine Learning for free at https://www.netsuite.com/twist
(21:14) Show Continues…
(24:36) Interview #2: Aaron Tan of Syncere AI
(25:13) Behind-the-scenes of that viral Lume "robot lamp folding laundry" video
(30:23) Stripe Startups - Stripe Startups offers early-stage, venture-backed startups access to Stripe fee credits and more. Apply today on stripe.com/startups.
(31:25) Moving beyond laundry: what else these Lume robots do?
Subscribe to the TWiST500 newsletter: https://ticker.thisweekinstartups.com
Check out the TWIST500: https://www.twist500.com
Subscribe to This Week in Startups on Apple: https://rb.gy/v19fcp
Follow Lon:
X: https://x.com/lons
Follow Alex:
X: https://x.com/alex
LinkedIn: https://www.linkedin.com/in/alexwilhelm
Follow Jason:
X: https://twitter.com/Jason
LinkedIn: https://www.linkedin.com/in/jasoncalacanis
Thank you to our partners:
(09:48) OpenPhone - Streamline and scale your customer communications with OpenPhone. Get 20% off your first 6 months at https://www.openphone.com/twist
(20:01) Netsuite - Download the ebook CFO's Guide to AI and Machine Learning for free at https://www.netsuite.com/twist
(30:23) Stripe Startups - Stripe Startups offers early-stage, venture-backed startups access to Stripe fee credits and more. Apply today on stripe.com/startups.
Great TWIST interviews: Will Guidara, Eoghan McCabe, Steve Huffman, Brian Chesky, Bob Moesta, Aaron Levie, Sophia Amoruso, Reid Hoffman, Frank Slootman, Billy McFarland
Check out Jason's suite of newsletters: https://substack.com/@calacanis
Follow TWiST:
Twitter: https://twitter.com/TWiStartups
YouTube: https://www.youtube.com/thisweekin
Instagram: https://www.instagram.com/thisweekinstartups
TikTok: https://www.tiktok.com/@thisweekinstartups
Substack: https://twistartups.substack.com
Subscribe to the Founder University Podcast: https://www.youtube.com/@founderuniversity1916
Transcript Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.
0:00 Speaker 00 Mirage is the only realtime video model in existence today. Even the big tech still don't have it, except for Google. Google does have this. But even the other big tech, as far as we know, don't have this. And it's challenging to build because you have to build a completely different model on the algorithmic side. You're training actually a different model. It's lots of algorithmic mathematical challenges. And at the same time, you have to write a bunch of assembly code for GPUs. Like, all this was written like in PTX, it was assembly for GPUs to actually get it to execute like this efficiently.
0:30 Speaker 01 This week in startups is brought to you by Stripe startups. Stripe startups offers early-stage venture-backed startups access to Stripe fee credits, expert insights, and a focused community of builders. Apply today on stripe.com/startups. NetSuite: The business landscape is very chaotic right now. That's why you need NetSuite by Oracle. Download the CFO's Guide to AI and Machine Learning for free at netsuite.com/twist. And OpenPhone: Create business phone numbers for you and your team that work through an app on your smartphone or desktop.
1:10 Speaker 01 Twist listeners can get an extra 20% off any plan for your first six months at openphone.com/twist. Hey, everybody. Welcome back to Twist. This is Alex. And today we have two absolutely amazing interviews for you. First, we're talking to the folks over at Descartes AI. You may recall that they recently raised 100 million at a $3.1 billion valuation. Quite a lot of money. What are they building? Well, Descartes got its start actually working on GPU optimization. And since then, has been working on world models. This comes in the form of Mirage and Oasis. And if that does strike a bell, that's because I showed off Descartes Oasis world models on the show last November. I've been dying to talk to the company about what they're building and where AI is going.
1:54 Speaker 01 So stay tuned. It's an amazing chat. Then after that, Jason and I are talking to the folks behind Sincere AI. Don't you know the company? Well, you actually do. They're the company behind that viral video of the lamps that sit next to your bed and then turn into robot arms and then do your laundry for you. Well, we were really curious, how far along is that product? What are the margins? What's it gonna cost? When's it coming? And also, what are they gonna build next? So stay tuned, because we have all that and quite a lot more from Sincere AI. So first Descartes, then Sincere. You're welcome. Let's go.
2:30 Speaker 01 Back in November here on Twist, we took a look at a really cool product from a company called Descartes. It had built a world model that generated a game around you. So you play the game and then it was kind of making it as you went. I was enraptured by this, just blown away. Now, I'm happy to report that since then Descartes has gone on to raise $100 million at a higher valuation and has stayed very, very busy shipping new goodies. So to help us understand what Descartes is building and why, and to explain what the hell a world model really is, please welcome to the show, it's Dean Leitersdorf. Dean, how you doing?
3:04 Speaker 00 Hey, how's it going? Thanks so much for having me.
3:07 Speaker 01 I'm so glad. Also, you're calling in from Tel Aviv where it's midnight. So you get like double extra credit bonus points for burning the midnight oil here.
3:14 Speaker 00 Midnight is like regular working hours. If I, you know, show you around the office right now, the office is still full.
3:20 Speaker 01 Okay, jokes aside, Descartes does a couple of things and I wanted to talk about each one quickly in sequence. You guys got your start doing GPU optimization, essentially helping people that are training AI models or running inference to dramatically lower their cost. And if I understand, Dean, that's where the company started, yeah?
3:38 Speaker 00 The company started by making a lot of money and that's a good thing to do. Okay, and the way we did it was by saving a lot of people lots of money. You really are good at getting the most out of GPUs and GPUs are expensive because Nvidia makes a lot of money. There's a handful of products with insanely high margins that make over $100 billion a year. The H100 is one of them. By the way, there's only four others, one of them being the Vatican. Very interesting, but completely different discussion.
4:02 Speaker 01 Similarly hard to get your hands on though. You can't really get the Vatican and you can't get a lot of H100s, there you go.
4:06 Speaker 00 But you know, next startup will optimize the Vatican. We, you know, we started off the cart, the cart's only less than two years old. First year was really focused on how do we get the most out of GPUs? How do we get them to be insanely fast, insanely efficient? From a software level, when we, when we started at Descartes, we're idling for a long time. Okay, what do we build? Do we build, you know, everything from a solar panel company to cybersecurity, to AI for healthcare? And, and just, just nothing stuck. But we were honest with ourselves and we realized that just once in our life. We gotta try to build a kilo corn. So you know, you have the unicorns, Deca corns, Centicorns, and kilo corns. Build a kilo is a thousand, build a trillion dollar company. Actually try to build
4:48 Speaker 00 a company that matters. Okay, build a Facebook, build a Google, build an Apple, build a company that if you look at the world before and after, it's a completely different world. Now, when you start thing like that, everything starts making sense. You don't have too many options. There's only a handful of things you can actually do. And in our DNA, I think there's pretty much only one thing you can actually do. Create a billion user consumer experience. That ChatGPT is the last billion user app out of it. I just can't.
5:17 Speaker 01 Oh, no, I don't think it is. But so billion user consumer application, it has to be something then that I'm going to touch every day. It has to be accessible via standard communication devices because you can't reach a billion people otherwise. So how did you narrow down to what to build to meet that standard?
5:37 Speaker 00 Okay, um, yeah, I can actually show you something. Let me, let me show you something.
5:41 Speaker 01 Okay, let's do it.
5:42 Speaker 00 Okay, I don't know if you saw this, but we launched Mirage recently. As you can see, this is Mirage, and you can see the real me, and now you can see me in the Versailles Palace, or you can see me in a blocky kind of world, okay? Or you can see me in anime.
5:59 Speaker 01 Oh, I don't think that suits you, actually.
6:01 Speaker 00 What about, oh, we can do this. This is one of my favorite ones. Okay, this is a wizard's world. Um, let me, let me find, one second, I'll get, here's a pen.
6:11 Speaker 01 Yeah, yeah, yeah.
6:12 Speaker 00 Okay.
6:12 Speaker 01 Okay.
6:13 Speaker 00 See this pen, see this red pen? Great. If I put it in the wizard's world, sometimes it becomes a magic wand, you can like cast magic spells with it. Okay, I could also change to,
6:27 Speaker 00 say, portal, okay?
6:31 Speaker 00 And if I have a shampoo bottle here.
6:35 Speaker 01 Standard office accoutrement.
6:37 Speaker 00 We actually do have a shower here because I am going to go shower after this before we go to bed because they have a bed in the office. But you know, the shampoo can turn into a portal gun. Okay. And you can you can try to launch portals with this every once in a while. It does work with the next version model. You'll be able to just say like, okay, launch portal and then like boom, it'll open up.
6:57 Speaker 01 So so Dean, let's talk about what this is. So this is a real time model that can take a video that you give it and automatically transpose it to a different style. But you don't have to give it a video, wait for it to process. It's happening as you're using it, which I presume was technically very easy.
7:16 Speaker 00 AI video models to date have always taken a lot of time to generate their their outputs. You know, you give it a prompt and then you wait a bunch of time and 10 seconds, maybe 20 seconds for some model, even a minute, and you get something back. Mirage has a 40 millisecond delay, like 0.04 seconds. Okay, so the version that you're seeing right now, you can actually play with is it takes in any input video stream, any prompt and just changes it based on the prompt. What we're going to launch pretty soon is also just, you know, generic video generation that's 40 milliseconds in. So you just tell it something and boom, it creates it.
7:50 Speaker 01 So in the case of the the real time usage, what you just showed me with it, making you into different forms, 40 milliseconds, is that a lot or a little? Because I know that if I'm playing a first person shooter in a competitive environment, that's probably a lot, but for pretty much everything else, that seems quick enough, I think.
8:08 Speaker 00 Exactly. We're also going to be dropping it pretty soon. It'll reach, it should reach like 16 milliseconds a month from now.
8:16 Speaker 01 Okay, and going back to the first topic of GPU optimization, I am just presuming that this does not melt entire data centers.
8:23 Speaker 00 Mirage is the only real time video model in existence to date. Even the big tech still don't have it, except for Google. Google does have this, but but even the other big tech, as far as we know, don't have this. And it's challenging to build because you have to build a completely different model on the algorithmic side. You're training actually a different model. It's lots of algorithmic mathematical challenges. And at the same time, you have to write a bunch of assembly code for GPUs. Like all this was written like in in PTX, it was assembly for GPUs to actually get it to execute like this efficiently.
8:53 Speaker 01 Okay, so you have to do the hard work of making the AI model and then you have to go essentially reprogram the GPUs to actually be efficient. So technically very impressive and visually, because I love this sort of thing, stunning to me. But I'm curious where you see it finding commercial application. You do have an example up on the site of someone looking down their sniper rifle site inside of a game. And I think there's an effect inside of there, which again, visually stunning, but is that where you see this kind of slotting into the economy today?
9:22 Speaker 00 I think it completely take over like half the internet. And I'll explain. What was the previous internet pre chat GPT? It was basically four things. You had knowledge, which was basically everything under Google's domain. Second category that you had was creativity. What do you do when you're not doing your homework? Okay. And that's, you know, TikTok, Instagram, YouTube. Then you had communication, WhatsApp, iMessage, and you had shopping.
9:48 Jason Calacanis? At launch, we talk a lot about systems over goals, right? You need to build a system, a process that lets your whole team hit those goals systematically. That's why your service business needs OpenPhone. You're going to get calls, you're going to get texts, you're going to get messages 24/7, 365 days a year. And with OpenPhone, all those random messages are going to be centralized so they're easy to find. And your whole team can share just one number, creating a truly collaborative system. That means faster response times. Plus, OpenPhone just added Sona, their AI-powered 24/7 customer service specialist. Your business is available around the clock. And because Sona keeps such accurate call summaries and records, well,
10:31 Jason Calacanis? human employees will have no problem picking off where the AI agent left off. So if your team needs to get organized, it's time to see why 60,000 plus businesses, including mine, are obsessed with OpenPhone. OpenPhone is offering Twist listeners 20% off their first six months at openphone.com/twist. What a great deal. And if you got an existing number with another service, OpenPhone will port them over at no extra charge. OpenPhone. No missed calls, no missed customers.
11:01 Speaker 00 Chatbots have completely taken over the first portion of the internet. It's gone.
11:05 Speaker 01 The knowledge portion.
11:06 Speaker 00 The knowledge portion, exactly. I don't know who will win, if it'll be OpenAI, or Gemini, or XAI, whoever. But chatbots are the way humanity is going to access knowledge and going to create knowledge going forward. That's it. All that is done. Shopping, I think, will be split in two. You're not going to browse amazon.com or Shopify. Your chatbot's going to tell you, hey, look, here's something cool, or you know, try this on, or whatever. Messaging, completely different topic, I can get into it, but I do think chatbots will take over that as well. And we're left over with creativity.
11:37 Speaker 01 Which is a quarter of the prior internet going by one of four categories.
11:41 Speaker 00 It is, it is the largest. And the nice thing is that as we get more productive and we spend less time doing our work, it's the only category that's growing, that fundamentally has to grow. If we now have to do four weeks, four, uh, four days a week of work, or three days a week of work, as you know, bots start, you know, being, making us more productive, we're going to have more and more and more creativity time. What's going to completely change your experience when you're having fun? That is, I think, the biggest open question today and consumer, and the consumer internet. Someone is going to answer that question in the next year. And whoever answers that question is, is going to be a giant company.
12:19 Speaker 01 So to me, what you're describing sounds like a new world that we inhabit that is persistently drawn by, I mean, perhaps something along the lines of your Oasis model.
12:30 Speaker 00 Exactly. And that's, you know, that's where we get to, to what a world model is. It has an incredible ability, just look at a bunch of videos and get how the world behind them works. And then just, just let you come up with completely new worlds. For the first time, we can take what's in our minds and just instantly show it to ourselves and to others. And really have a completely immersive experience.
12:54 Speaker 01 No, I love that. Now, what you're describing to me still feels relatively single-player, because the entertainment or consumption portion of the inner that we described, Netflix and TikTok and so forth, mostly consumption, not as much creation. It's a single-player experience. What I think you're kind of talking about, though, feels much more multiplayer, as in you and I are doing this together. So are you and I sharing down the road when we're working towards a billion people in a kilo corner, whatever the fuck it was? Are we together or are we still a part in that world?
13:28 Speaker 00 You know, something mind. So, you know, we just put up the Mirage demo three weeks ago, and we've been mind-blown by what people do with it. And what blew us the most was that people actually did physical experiences together. So we had, we had so many people just took out like broomsticks and started hitting each other. And they put like a Star Wars prompt and it just turned into lightsabers and hitting each other broomstick. Suddenly becomes, you know, an okay thing to do when, when, when the world sees it as you guys having a lightsaber fight. I really think it'd be mind-blowing to see if it creates a new reason for people to hang out in person. And, and just have some physical combined joint interaction and have the AI cast that into a different world.
14:12 Speaker 01 Did you ever play with Microsoft's HoloLens headset?
14:16 Speaker 00 No. We do, we do have the Quest 3 and the Vision Pro here, but not the HoloLens.
14:21 Speaker 01 HoloLens blew me away. So I used to be a Microsoft beat reporter. So I just covered the company. And I got invited up to Redmond for the HoloLens first day. Like, no one knew this thing existed. And I was playing, I put it on, and suddenly I could play a game that was on a tabletop that was real, but the assets were digitally drawn on top of it. And I like, you know, pushed a little, it was kind of a Minecraft-ish game, like everything is at that level. And I kind of like pushed blocks through tables and got to interact digitally and physically at the same time. And I've never quite forgotten how straight-up magic that felt. So I do think there's a point of synthesis. I think it's actually more of a physical form factor question than it is how does the software work?
15:00 Speaker 00 100%. I think it's a physical form factor. Something that blew our minds with Mirage. Do you know how many hotels reached out to us? Hotels! Hotels reached out to us and said, "Can we stick this in our lobbies?"
15:14 Speaker 01 Oh, as a game?
15:15 Speaker 00 Exactly, as like a new game, a new experience. They're like, "Can we just stick this in our lobbies?" They will put up a big TV screen and a nice camera on top, and people will just walk by and be like, "Huh, I'm in Frozen right now." And use that as a gateway to your arrived at your vacation. Imagine like you're going to a resort in the Bahamas, and so you walk up in front of the screen and you're all sweaty after your flight and everything, and suddenly, boom! You're in your beach shorts and everything's sunny. And we're like, huh. We never imagined that this is what people want to do the first week in Mirage.
15:54 Speaker 01 That is surprising, but if I was a Las Vegas hotel, I would put you in a James Bond-style tux with a martini and a sack of roulette chips, for example. Or if I was in Russia, maybe a funny hat, or whatever it is. I like that, but there's several steps between where we are today with Mirage and Oasis. I can see them coming together. I can see the vision. I also think that you and I have a similar optimistic view of where we're going. I guess what I'm curious about is, what are the steps in between? And are they frequent and small, or infrequent and large?
16:29 Speaker 00 It took us seven months between Oasis and Mirage, because there were lots of key breakthroughs that we had to solve. I think that a year from now, if we have this interview again in a year, we'll be living in a completely different world. And the answer of who gets to be the AI that you use when you're not doing your homework will probably be answered in this upcoming year, just because there's so many product advances that are coming out based on this tech.
16:53 Speaker 01 Now, based on this tech, are you referring to AI as a whole, or what you guys are cooking up in particular?
16:58 Speaker 00 AI video and audio. Combination of AI video/audio models. Real-time is just going to be incredibly fast, and it's going to also have so many products built on top of it.
17:11 Speaker 01 So this is a business show. I know it doesn't sound like one so far, but when people talk about building models, usually what I hear in the background is the crackling fire of entire bales of cash. You guys told Fortune that you had burned incredibly small amounts of money, like 10 million when you raised your last round. And usually when I talk to a company that's doing that, they found some way to have a very efficient go-to-market motion, or whatever it is. You guys are doing the hard, expensive thing and not losing money. And I'm just kind of curious, is that going to be something you can keep doing through this process of progress? Or is it gonna get more expensive to fulfill this vision?
17:49 Speaker 00 What's been very central to the card from day one was being able to actually utilize GPUs incredibly efficiently. And we're significantly better than that in the rest of the industry. By doing this, we're actually able to do what costs hundreds of millions of dollars for other people in order to make through less cash. That's a key fundamental advantage that we have is based on the tech stack that we built over the past two years. It is surprisingly hard to replicate. Now that we have this as our bedrock, it really sets us apart in being just able to operate incredibly more cash efficiently and create new products that other people just can't create.
18:28 Speaker 01 Okay, so you have the right team. You can make this stuff cheap. Things are moving quickly. I guess then the right next question is, what is coming? You've hinted now at new models that are coming in the next couple of weeks or months. So without breaking your own internal NDA's, what is the, what should I be looking for from you guys as you work towards this trillion-dollar, billion-user future?
18:49 Speaker 00 So you're gonna see three things.
18:51 Speaker 01 Okay.
18:52 Speaker 00 From the lab, from the experiences team, and from the B2B team, from the enterprise team. From the lab, you're just gonna see insanely cool models coming out. I can tease one of them already. Imagine in real time, now just Mirage currently changes the entire world. The next version also gets to make local edits. So I can, for example, you know, just be like, hey, give me a sword and then boom, I get a sword without anything else changing. Uh, give me, you know, a long ponytail, then boom, I get a long ponytail and, and you know, like make it like ninja style hair and make it like sleek black or something and change my entire clothing to ninja.
19:28 Speaker 01 You're going to turn every Zoom into an RPG character creator screen?
19:33 Speaker 00 Exactly. The amount of products you can build on top of this is insane. We're going to build our own like Mirage Zoom, Mirage FaceTime, Mirage Snapchat, Mirage YouTube, mirror everything they currently imagine just with, hey, let me just change the entire stream in real time.
19:52 Speaker 01 So is that one app? Cause you could do, you could do Zoom and WhatsApp and everything else as a combined experience that is multi-platform or as a multi-part product? So what are you thinking?
20:04 Jason Calacanis? Despite what they may say, no one has a crystal ball. Nobody can predict the future and even me, no Shrek on us. I make mistakes now and again, but what you can do is future-proof your business with NetSuite by Oracle. NetSuite is the number one enterprise resource planning tool, ERP on the market. What's ERP? Well, it's going to bring together every facet of your business into a single unified platform, accounting, financial management, inventory, everything in one place, giving you the control and visibility you need to make really informed decisions. And you can't strategize and forecast anything about your business without the most up-to-date and accurate information possible. At launch, we are huge sticklers for data.
20:46 Jason Calacanis? Every Monday, the team meets, we review their goals, their strategies, and most importantly, they share specifics on how everything performed. You spend a little time looking backwards and you can accurately and confidently focus on what's next. And that's what NetSuite by Oracle is going to help you do. Our friends at Oracle want to give Twist listeners a free gift. Download the CFO's Guide to AI and Machine Learning for free at netsuite.com/twist. Once again, netsuite.com/twist.
21:14 Speaker 00 We're going to be launching this in a combination. So a lot of it's gonna be combined to one, some of it'll be standoff. Uh, for example, gaming, gaming makes sense as its own gaming platform. Uh, we're gonna be launching mods for some of the popular games. For example, you can install a mod and boom, change your Minecraft world into Frozen. So we're going to, we're, we're gonna see gaming platforms, we're gonna see social platforms. We have lots of products that we're gonna launch very soon surrounding, built on the, on this tech that we're creating. And last part is, we just had so much inbound for enterprise usage for this. From the entertainment space, from the education space, by the way, teachers love this. They can, you know, just turn their entire classroom into safari or to dinosaurs or whatever.
21:59 Speaker 01 School has changed since I was there.
22:01 Speaker 00 Teachers are looking for something to keep kids, kids engaged. Um, you know, something mind blowing that happened, hospitals reached out. One hospital reached out and was like, you know, a big problem with kids' hospitals is that, you know, when you do procedure on a child, they're, they're scared and, you know, they're, they're uncomfortable. And so they try to distract them and they showed them Mirage and the kids were just like captivated by it. Like, holy shit, look at what happened to me. I turned into something super cool. And, and the hospitals, the doctor was able to do whatever they need.
22:33 Speaker 01 Dentist office, my, my spouse is a, is a physician who is also a pediatrician. So this is very apropos to, to the world that I'm right next to. So I can see the absolute demand for that.
22:46 Speaker 00 So in different domains that we probably can't even imagine where it begins, but where it ends, but we can, we can see where it begins. And, and so you're gonna see all three things, you're gonna see new models, new consumer experiences and new enterprise applications. And then it'll be, it'll be quite a ride. We have, we have a few amazing weeks coming up.
23:06 Speaker 01 Okay, Dean, we've already gone way over time, so I, I have to kind of wrap us up here. But, uh, being very honest, I saw what you guys built. I loved it. I saw that you raised a hundred million, I think like a 3.1 billion dollar valuation and I went, huh. Now I get it. There aren't that many companies, apart from maybe OpenAI, that are thinking this expansive and long term. And if you're right, cool. And if you're not, cool. But I mean, at least you're taking a swing big enough that you might smack the moon. So I'm here for it.
23:37 Speaker 00 Once in your life, you gotta give it a shot. Either it works, or you had the best ride you can possibly have.
23:44 Speaker 01 All right. Tell people where they can find Descartes on the internet. And if you have a role you're having a hard time filling, shout it out so the world can gather around.
23:53 Speaker 00 So we would love for you guys to visit Descartes.ai or our Twitter accounts, TikTok accounts, all our socials. The best role that we have to fill right now, we have insanely talented people. We the most talented, young, marketing person possible. And you can't have a nice marketing degree. What you need to have is insanely creative TikTok or Instagram or YouTube short skills to just be able to take this visual tech that we're building and show the world what it can do.
24:25 Speaker 01 All right. Well, Dean, thank you so much. And we'll have you back not in a year. It'll be sooner than that. Cause I'm sure there's gonna be a lot to talk about, but I appreciate you very much. Go have that shower, get some rest, man.
24:33 Speaker 00 Talk soon. Good night.
24:37 Speaker 01 A couple of days ago, there was a big viral video on Twitter of a couple of lamps that turned into robotic arms and began to fold clothing. Everybody hates doing laundry. Everyone loved this idea. Many questions about what it can be used for and where it is. But to help us explain all that and more, let's welcome Aaron Tan from Sincere AI, the company behind the Loom robot lamp. Aaron, hey.
25:00 Speaker 03 Hey, how's it going? Thanks for having me.
25:02 Jason Calacanis? I saw the video and I thought immediately, and we'll play the video now while I'm talking. Number one, this is AI generated. It's not real. It looks too insane to be real. Please confirm with me that what we've seen here is a real video.
25:24 Speaker 03 It's not an AI-generated video, but it's also not a real video either.
25:28 Jason Calacanis? Oh, okay.
25:28 Speaker 03 It's a render of our design. Yes.
25:32 Jason Calacanis? Okay, so this is a render. That's because it felt a little like too perfect. So you're making these lamps with these precision clips on them we're playing right now. Explain to me why you chose, as an entrepreneur, this specific design. And these calipers at the end, if I could call them calipers, they look like calipers, pinchers. And you have to have an initial use case. You went with folding laundry. And with this specific design, it's iconic, it's quixotic. Why this?
26:07 Speaker 03 Right, yeah. I mean, when we started this company, our main goal was to get rid of chores for people. So we wanted to build robots to automate those things away, right? But as we, you know, we started at the beginning of the year, and as we worked on this idea of just like robots to automate chores. The thing that we learned over the time was that nobody actually wakes up and asks for a robot. People just want a way to get rid of their chores. And no one really wants their home to look like a factory with these free-roaming industrial-grade machines. That poses kind of like safety and privacy issues. So essentially, like, one thing led to another, and ultimately, we came up with this idea where it was like...
26:51 Speaker 03 And thanks to Beauty and the Beast as well, that was one of the main inspirations.
26:56 Jason Calacanis? Oh, right. Beauty and the Beast had like the candelabra, the teacup.
27:00 Speaker 03 Yes.
27:01 Jason Calacanis? Amazing. Wow.
27:02 Speaker 03 Yeah, yeah. So one thing led to another, and we basically just like came to the realization that, you know, we want to get rid of chores for people, but we want to do it in a way where it doesn't feel like we're adding robots to your home.
27:13 Jason Calacanis? Got it.
27:13 Speaker 03 And the best way to do that is to hide robotics in plain sight and have them work in this sort of ambient background manner where you don't realize there's anything really there.
27:22 Jason Calacanis? That's why they, when they're at rest, they're just gorgeous, well-designed lamps. Got it.
27:29 Speaker 03 Right.
27:30 Jason Calacanis? When will you deliver this? And at what approximate cost? Because I think you're taking pre-orders.
27:36 Speaker 03 Yes, yes. So we have a preorder list now. The majority of the people on the waitlist will get theirs 12 months from now, but we've sort of started this new like early beta tester club. And we're gonna get, we're gonna get the units of those folks within six months.
27:51 Jason Calacanis? Amazing. Proximate cost for a pair of these? Yeah. Everybody wants to know.
27:56 Speaker 03 Price is somewhere between one to $2,000.
27:59 Jason Calacanis? Got it. So this is an early adopter type thing. If you were paying somebody to fold clothes for five hours a week at $30 an hour, that's $150. That would probably be a family's clothes. So if I were to look at it to get rid of that chore, if I was paying somebody to do it, which is the people in this benchmark, Alex. 150 bucks a week savings for five hours of laundry, which I think is what a family of five might do.
28:28 Speaker 01 Yep.
28:29 Jason Calacanis? Uh, pays for itself in 10 weeks. Is, is that about what you're thinking, Aaron?
28:34 Speaker 03 Yep. That's, uh, that's, yeah, exactly.
28:37 Speaker 01 How is it so cheap? That I, I thought it was gonna cost me five and then I was gonna have to do the math Jason just did. But if it's $1,000, I, I kinda feel like it's almost an impulse buy.
28:46 Speaker 03 Yeah, yeah. Well, we're trying to push it as sub 2,000 is what I've said on the internet. We're trying to get it as low as possible. Um, the first batch is probably going to be closer to $2,000. Um, yeah.
28:58 Jason Calacanis? But the bomb on something like this, I'm guessing is 80% of that cost, right? This isn't super expensive, but the first iterations you'll break even at. While you refine it is a kind of as an entrepreneur of the plan?
29:11 Speaker 03 Yes, yes.
29:12 Jason Calacanis? Okay. So, go ahead, Alex.
29:14 Speaker 01 Oh, I was just curious, you know, what are you guys gonna work on next? Because I have a list of chores that I hate doing and I could write it down for you. It's quite long. Um, so once you figure out the laundry inside of this, I presume since you're AI, your company is not going to stop at just folding stuff in my bedroom. So what's the next robot that's going to come out? And what, what are you gonna hide it inside of?
29:34 Speaker 03 Yeah, I think, I think as of right now, the current plan, at least what I'm allowed to share is that we, we want to keep to the same form factor. Um, but it will do more than laundry folding. Um, we, we sort of envision sort of, these are like single floor lamps that you can kind of place anywhere in the home. They have limited mobility, so they can move around a little bit. Um, and the whole idea is that depending on where they're placed, uh, they can do tasks that contextually makes sense in that particular area. Okay. So when it's like in your bedroom, laundry folding is, I think, everyone can agree is like the killer feature. Um, but there's like other little things that people have asked for, like bed making or, you know, even just simple things like holding an iPad in bed uh for you,
30:18 Speaker 03 all the way up to like massages and back rubs and, you know, all that kind of stuff, right?
30:25 Jason Calacanis? Everything is moving so fast at early-stage startups today. It's amazing what I'm seeing people get done. Founders need simple solutions that can live in the background and solve their problems, right? Well, now there's Stripe Startups. Their startup program now offers a ton of benefits for your business. And this is when you know, when a business like Stripe specifically creates something for startups, you know, they really care about founders. So you get Stripe-free credits, you get a unified dashboard where you can track your performance, and you can even start issuing your brand's own payment and credit cards in just weeks, not months. For AI companies specifically, that market is shifting and changing every day, requiring an unprecedented level of flexibility from payment providers and partners.
31:07 Jason Calacanis? That's why 78% of the Forbes AI 50 are already using Stripe as their financial infrastructure. Here's your call to action. Just learn more about this incredible program and apply today by visiting stripe.com/startups. Very simple, stripe.com/startups.
31:25 Speaker 01 Yeah, there, there's an obvious next question to the back rub massage element of this. And you guys have, have spoken that those adult quote-quote use cases are, are off the road now, right?
31:36 Speaker 03 Well, I'll say this, we'll have an app store. Um, and the app store.
31:40 Jason Calacanis? Here we go. You could get interesting quick.
31:44 Speaker 01 I couldn't, I could not have served up a bigger softball for you and you just said, no, I'm not gonna swing at that.
31:48 Jason Calacanis? I wasn't even, I didn't even consider your light of questioning, Alex. Well done.
31:54 Speaker 01 Oh, wow. Yeah, no, I, it's got, it's grabby hands in bed, Jason. I mean, there's, there's, there's things that people are gonna do with it. But, um, uh, people have said that, uh, you actually said that there's gonna be over the air updates to this thing. So it's gonna get smarter after I buy it. If I get the lamp from my bedroom, can I take it to my kitchen later on and have it help with the dishes? Or will it be a separate device?
32:15 Speaker 03 You can certainly carry it and place it there if you like. Um, but I, I sort of see this future where, uh, people probably have multiple floor lamps, uh, in their homes, uh, all different, doing different things. Um, yeah.
32:29 Jason Calacanis? I mean, these, the way it's designed, you could easily see it by a kitchen island.
32:33 Speaker 03 Exactly.
32:34 Jason Calacanis? Cutting vegetables. Does it have casters on the bottom where it will move around your space? Like a caster's fancy word for wheels, I guess.
32:43 Speaker 03 Yeah, yeah, it does have wheels and they are actuated. Uh, that's how we have it in prototype. I think the main, the main thing we're trying to make sure we're doing is that we want to limit how much range of motion it has for safety reasons. I think that everybody in the home should at all times, um, know where the robot exactly is. So there's no sort of like free roaming upstairs, downstairs into your private space, all that kind of stuff. So wherever you put it, whatever contextually makes sense in that limited range, we're picturing like a meter, two meters tops. Um, it can handle the task there.
33:16 Speaker 01 How does it see? Cause to be doing laundry, it has to be able to have vision and to do other things. But also you mentioned privacy and I just realized that if I do have both these lamps in my bedroom, I might not want them to be washing all the time.
33:28 Speaker 03 Yeah, yeah, yeah. No, that's a great question. This is something that we thought carefully of. Uh, the best way for me to show you is, um, we have a. Sort of a early prototype here. So this thing is designed, the short answer is that the camera and everything that is robotic about this lamp is housed inside the lamp hoods. So only when you give it permission, does it then open up, reveal the camera, reveal the robot fingers. So these these lamp hoods almost act as, not almost, they act exactly as mechanical shutters that some folks have on their on their laptops or whatever. So only until you've given it permission, does everything robotic about it reveals itself. Otherwise, it's all nicely packaged and sort of contained inside lamp hoods.
34:13 Jason Calacanis? Genius. How much did you raise to, or does one need to raise? And what lunatic backed this company? Because this is, I mean, you gotta have a real, yeah, this takes a unique person, I think.
34:30 Speaker 03 Yeah, yeah, it's a bit unorthodox.
34:33 Jason Calacanis? Unorthodox, thank you.
34:34 Speaker 03 Yeah, I've, you know, I've heard all sorts of things. You know, when we, when we first started pitching this, we were like, oh, we're gonna build robot lamps that full laundry. And everybody sort of told us like, those words don't feel like they belong next to each other. But in terms of the raise, we're actually still in the middle of the fundraise, hence the reschedule of the of the pod. I would have more to share on that maybe a week from now.
34:59 Speaker 01 Okay, well, I'm stoked, Jason. I think this is awesome. I think this is exactly what I want to see. I think we've talked about humanoid robots in the show a lot because they're a great idea for a lot of settings. But I also think it's a good point about the house. I don't actually probably have room for a humanoid robot to be bouncing around the house. We get in the way. So.
35:17 Jason Calacanis? That's what closets for. There's gonna be like a, I think everybody's gonna have a robot closet where the robot goes to just, you know, tell it to go into the pantry. And it's like C3PO is gonna be in the pantry like, oh, what did I do? It's like, just go go into the oil tank. Remember, he goes into the oil tank and does all this stuff.
35:38 Speaker 01 Have you read Murderbot, Jason, the series?
35:40 Jason Calacanis? I have not.
35:41 Speaker 01 Murderbot is about a robot that's a humanoid robot, and it's it's it's just kind of sad all the time. And so it's this interesting story about like, it's basically a bodyguard robot that has sentience and gets sad. And it's now a series on, I think, Apple TV plus. But I think I think we're gonna end up with very similar. He has a closet he goes to repair himself, which is why it came to mind. But I think we're going that way, too.
36:03 Jason Calacanis? Aaron, this is incredible.
36:05 Speaker 03 Thank you.
36:06 Jason Calacanis? What's next? This is obviously one of a collection of robot operating system you're building in App Store. So you're going to have four or five different SKUs, I take it, over time?
36:16 Speaker 03 Yeah, yeah, I mean, I think there's, I mean, lamps is, lamps are such a sort of a personal thing that I think people probably have different styles that they want that fits
36:25 Jason Calacanis? Okay, sure.
36:26 Speaker 03 different parts of their homes. And we sort of see a product lineup of everything from tabletop lamps to floor lamps, to lamps with multiple heads that could do different things. Yeah.
36:38 Jason Calacanis? That's the thing I think would be interesting because you have those pinchers. But there are other attachments. So having it be able to go to, I don't know, maybe it's in its base, have five different types of hands for different purposes. Obviously, you don't want to be massaged with
36:57 Speaker 03 Exactly.
36:57 Jason Calacanis? pinchers that could be, I mean, unless you're into some Shiatsu, I guess, or if you're into acupuncture, but whatever, whatever your bag is, Alex, you know, maybe you want pain in your massage.
37:08 Speaker 01 I'm sorry, I'm a really big science fiction fan. I'm pro robotics. I love technology. I'm a capitalist. I will never get acupuncture from the lamp beside my bed. That's where I draw, I found it. There's my line, Jason, right there.
37:20 Jason Calacanis? Wait, you're saying Murderbot editorial director Lon, air director, says Murderbot's going to become an Apple TV Plus series?
37:29 Speaker 01 Yeah, I think it's, I think the first season's out right now.
37:32 Jason Calacanis? What? All right, you learned something new here. All right, listen, Aaron, continued success. If you're raising, you have my email. If you happen to raise money at some point, I might want to, I might want to join this lunatic investor group that you're assembling, if you are in fact, or maybe you're just doing this off of profits, but, which is also notable, but, um, gosh,
37:54 Speaker 03 Yeah, yeah, we're closing soon.
37:55 Jason Calacanis? this is the kind of crazy I'm here for.
Previous Meta's "Sensual" AI Doc LEAKS, Opendoor's dreams became memes + OpenAI plans $1T data center spend | E2165 Next Scaling AI Photo Editing to 300M Users with Photoroom's Matt Rouif | AI Basics with Google Cloud Page updated November 1, 2026 · transcript and summary are machine-generated
