He sold a company for $1B+ to Adobe — Startup hacks from Emery Wells

Sup nerds, welcome to the Jason Levin Show where I talk to the craziest viral marketers blowing up the internet.

Crazy.

What are you waiting for? Let's go blow up the internet. Sup nerds, today I'm talking with Emery Wells, the founder of Frame.io, which he sold to Adobe for over $1 billion. Frame.io is a video review platform. It is the number one video review platform in the world used by creators like me to high-end film studios out there, the, the biggest and best use it. It's pretty sick. This is not a paid ad. I actually just love the app. Not gonna lie. It's pretty cool. So we talk about everything. So Emory started his career as a video producer running a boutique creative agency, working with companies like SNL in New York. And then he decided to go build software. And we talk about his journey going from agency life to building a software company, which he sold for over a billion dollars. Pretty cool stuff. A lot of lessons there around management and scaling. We also talk about what he is investing in now, wink wink, robots. Pretty sick. And we also talk about his predictions around AI and art. This was an awesome conversation. I learned so much from it. Like, after I filmed this conversation, I'm telling you, I just pulled out a pen and paper and was just writing down so many ideas. It was a lot of fun, and I hope you enjoyed this conversation as much as I enjoyed it. All right, Emory, it's been cool seeing you start blowing up on Twitter. Uh, been posting a lot of wisdom and I wanna talk and learn a bit more about your story. I know reading about you, moved to New York when you were just a teenager, right?

Yeah, I was 20, so a little one year past teenage years, but yeah, I was 20 when I moved to New York.

And getting started with filmmaking, right?

Yeah, I moved to New York City originally with the dream of becoming a filmmaker. Very classic. New York City story. I just moved here with $500 in a suitcase and was gonna figure out how to make movies. That was the original plan.

And you ended up creating shorts for SNL, correct?

Yeah. The first half of my career was in filmmaking and I wound up starting a, a company, a post-production company, and SNL was one of my clients. There were, was, I did a lot of television commercials, diaper commercials and all kinds of stuff, yogurt and just run-of-the-mill commercials. But SNL was one of the, one of the more fun clients I had. So I worked, they were a client for about 4 years and I did the post-production. My company did the post-production for over 100 of their digital shorts. Wow.

And how big was your company at the time?

It was tiny. It was a boutique agency. The typical model in, in the kind of creative agency world is the, this, the run-of-the-mill kind of company is a boutique creative agency. There's like a guy like me who is the principal and often even doing a lot of the creative work. And then most of them are probably like sub-20 people. Mine was 5, like 5 people. And then I would hire a bunch of freelancers for busier times. But I think the most, like the, yeah, I think the peak of the full-time employees I had, of course they weren't even really full-time. They were just like permalancers, but I considered them full-time employees. So it was small. It was very small.

Nice. Nice. But you decided to leave the agency role behind, go for something a little more scalable.

Yeah, I'd always wanted to do tech. I had been in post-production. Post-production is, when I say post-production, there's a lot of different aspects of post-production. There's a lot of subcategories within post-production, visual effects and motion graphics and color grading and finishing, all these different things. And post-production and especially the more technical side of post-production is very similar to software development in terms of it's this intersection of deeply technical and deeply creative, which is how I think about software development, especially as you get into like bigger scale visual effects and 3D workflows. Those companies look a lot like software companies. They have engineers, they have— really game companies are like the intersection of software and visual effects. So anyway, I felt like I was in that zone of, I loved the intersection of technology and creativity. And, but I was not really a— I did not have a computer science background. I did very lightweight coding for visual effects purposes, like scripting and things like that. But I always wanted to get into tech and, and really the catalyst after spending nearly 10 years doing post-production was when I met my co-founder, who was somebody I, I hired right outta college to join my post-production company. He had a computer science minor, and then together we started tinkering and we were the right combination of skills to do more serious software. We built a little iPhone app as our first project. This was like 2007, '08, something like that, when App Store was relatively new. And that was the first thing I'd ever— first software project I'd ever done was a little iPhone app we sold on the App Store. It was such a magical experience. I remember the first time you like— we got the app running on an actual phone because you work on it in the simulator. Then we figured out how to get it running on the phone. It was like, oh my God, this is— it seemed— it was such a fun and magical experience. We sold this little app on the App Store as a calculator. For $5 and we had so much fun doing it. We said, well, we have to build more software. And I felt like he and I had the right pairing to do it. And so we worked on a few things, but ultimately decided to work on what became Frame.io and then the rest is history, so to speak. Could chat more about that journey, but.

Yeah, dude, I'm going through something similar right now. So I've been making content since I was a little kid. I actually went to a VFX summer camp. ID Tech, if you— I'm sure you've heard of it. And so I was super into it, YouTube as a kid, and then all into blogging, writing. And that's where I really found my strength was writing online and was able to do that. And so I've been able to build up this very profitable business. But then there's that bug that starts itching you when you're around startups all day. Oh, why am I not building software? What the hell is wrong with me? If these guys could do it, I could do it. And so started playing around with Bubble, started screwing around, building stuff with a friend. And now that feeling of magic of seeing that working on your phone, like we, our first little prototype, it's a pretty, pretty special thing. And there's so much art that goes into it. The same creative mindset that goes into writing blogs or writing a book or designing on Photoshop goes into building software. And I think it's really underrated.

So yeah, definitely. And I know a lot of, I have a lot of friends that, that are in media in one way, one form or another. And, and. But when I say media, I mean that broadly, like you're making content. So many different, so many people make content, whether it's an agency or you're an actual content creator, like an influencer. But the challenge with those businesses is they're really hard to scale. And it's a, you're trading time for money. And what most gurus love to say is that you'll never become wealthy trading time for money, which is true. You have to have a product to sell and software is one product to sell. It's not the only product to sell, but it is very scalable.

Yeah. Yeah. I think you can make good money, but not like real wealth doing it. And I'm—

all relative, of course, all relative. Everyone has a different definition of that, but yes, it is harder to continue to scale those types of businesses.

Yeah. I'm hitting that point where I'm a little like, all right, let's figure out how to the next step. And so jumping in early days of Frame, you guys just launch, you're a total outsider of Silicon Valley, correct? If I understand.

Yeah. I didn't know a, I didn't know a soul in Silicon Valley. I didn't know an engineer. I didn't know a VC. I didn't, I was just, I did not know anybody. I didn't know somebody like you. I didn't know anyone that came from that world. Everyone I knew was in post-production and this was, I started working on Frame in like 2013, '14. We launched in 2015. So different era back then, but I really didn't know a soul.

Yeah. And. How did you break in?

We broke in, we launched, we did, we, when we, so we were building it for about 2 years before we just, me and him, just me and my co-founder, John. And we, when there was a time where we thought, oh, we're really close to actually being able to launch this. So let's have a little like public sort of pre-launch moment or announcement. And we thought we're going to announce it and then it's going to be ready shortly thereafter. Turns out that when we did announce it, it took us a full year to actually get it ready for people to use. It wasn't as— it was further out than we thought, but anyway, we did an announcement and I wrote a tweet about this, which was like everything we did to create virality around the launch, but we wound up having a, you know, what I would say is semi-viral successful launch where we got 15,000 people to, to sign up for, to be notified when they were going to be able to use it and the, The experience that we built had all these calls to action where we asked people to, we gamified the launch. It was like, hey, you want to get access to this new thing that we made look really sexy and cool? Do all these things on our behalf and you'll move up in the line. And so we asked you to share this tweet. We had a pre-written tweet, share this on Facebook, a pre-written post, follow us on social. So we created a flywheel and we did many things, but I think one of the most important things we did was we created this flywheel, which was an ex— I would honestly say we, we created a landing page that took you through an experience. Once you got on, once you entered your email address, most launches, let me back up. People make a landing page and of course there's a place to enter your email and that's where the journey ends, right? It's enter your email. Thanks. We created an experience around the, the after entering your email. And I think it's such a, it's such a people really miss this opportunity. When somebody enters an email address because they're curious about a product, you have their attention, right? They're engaged, they want something. And so it's— you should utilize that moment. And we— I would say that we put people on a little bit of a ride that was a combination of design and copy and animation that once you got on the ride, you didn't get off the ride. It was like, we're taking you through a series of steps. And it was so fluid and the right copy. You were just You just went through it. We had 15,000 people that went through this multi-step journey and we had all these asks along the way. And I really believe that through the right combination of copy, animation, and design, you can get people to do whatever you want. And we did that. And so we got 15,000 people to sign up and all this sort of virality on social and generally broad kind of engagement. It was actually a very, we were, I think on the super early distribution list of Product Hunt when it was like not even a website, it was just an email list. I didn't even know what it was. I found found out later that we, somebody put us on that email distribution list of Product Hunt. We were on the front page of Hacker News because for a completely unrelated reason to our product, we just happened to use this technology framework, a front-end framework called Famo.us, which was getting buzz at the time. It was like, it was a front-end, it was a front-end framework and we used it to build the landing page and front-end frameworks have always been, I suppose, but even at that era, it was like a very controversial. It was like, it was very controversial. So we used this frontend framework and the fact that it was like one of the first kind of real projects that used this experimental frontend framework, we were the top story on Hacker News. And so between the virality of the signup flow, Product Hunt, Hacker News, we got on the radar of some Silicon Valley folks, which was the very, very first person that I spoke, that I made contact with. Actually interesting. The two first people that I ever made contact with in Silicon Valley was one was Aaron Levy from Box. So Aaron Levy reached out to me via email because he saw it. He's, hey, I'm Aaron Levy, saw this thing you did. That's cool. And, and oh my gosh, why am I completely blanking on his name from the CTO at Coinbase? That was Adam Dreesen before that. Why am I just having like a brain fart moment?

Balaji?

Balaji. Yeah. Balaji.

Yeah.

And he was so the very first investor that made contact with me was Balaji. And the reason that he made contact with me is because one other element of our signup flow flow was after you went through this whole gamified experience, you got a timed email from me that was timed to go out exactly 15 minutes after you signed up. It was plain text. It had no unsubscribe. I really tried to make it look real and not automated. And it was a very conversational copy. Hey, and I, and it appealed to, the copy was something that sort of appealed to, I was like, oh, you signed up. I noticed you as an important person signed up. Can I ask you what interested you or whatever? Course, meanwhile, it was going out to absolutely everybody. And but I got thousands and thousands of responses. And I read all of those responses. And within those responses were some like real nuggets, like people like Balaji, which was the first contact I made with an investor. And I had a little email exchange with him. Turns out it wasn't the category wasn't really his wheelhouse. But he wound up introducing me to Steven Sinofsky, who was the former president of Windows at Microsoft and was now a board partner at Andreessen. So the very, very first people I made contact with in Silicon Valley happened to be some pretty serious people. There is a story, there's a story about Aaron Levy at Box where he did make a play to just acquihire us before we launched. And he was, but he was the first person I met from Silicon Valley, by the way, awesome dude, like really enjoyed my experience hanging out with him, getting to know him after he contacted me via email and I made contact with Balaji. I took to the couple of people I was talking to. I did. I was like, hey, I'm going to be out in the Bay Area. I wasn't going to be out in the Bay Area, but I'm going to be out in the Bay Area if you would love to connect. So I wound up having dinner with Aaron and a bunch of his executive team, which was such an awesome dinner. It was just like to be like a total outsider from that world. And Box and Aaron weren't exactly what Box and Aaron were now, but it was still, I knew him. I knew of him at the time. But to make contact with those folks very early on was really exciting. I didn't wind up doing any deals though with that, that contact with Balaji and Steven Sinofsky and Andreessen Horowitz. They wound up introducing me to some other people and, and I pitched a bunch. This was pre-launch. So I had that, we had this sort of viral launch, but the product wasn't available and it wasn't selling. So we were not actually able to raise money during that, from that pre-launch all the way till the actual launch. Like our, the feedback was, wow. Really awesome product. Love you guys. Come back to us when people are paying you money. And we were outsiders. So when you look at, if you're an out, for the people out there that who may be outsiders and you're reading the TechCrunch headlines and the whatever headlines about these companies that are getting funded, it seems like everybody gets money. And it seems like you can get money without having even a product. Like you're just like these rand— like, it seems like it's so random. It's not random. The people that get money that are like stealth mode, pre-launch, whatever, are 9 times out of 10 people who have previously had a successful exit or were known quantities. They were already had a company, or they're the right profile of ex-Facebook or whatever, ex-Meta, ex-Google. If you're like the ex-Google person leading an AI initiative, you can go off and have nothing and raise money right now. So it's all about the dynamics of the market, the story of your background. My story was I never done software. I was a post-production guy. We'd built this app that we could show people and demo, but other than that, I was an unknown quantity. So we pitched a bunch during that year between announcement and actual launch and couldn't get anyone to lead. We got a lot of, hey, we'll join in if you can find somebody that wants to lead the deal, but nobody wanted to lead the deal, which is very common. People, there's a responsibility you take when you lead a round that's very different than following on a round that somebody else is leading. I didn't understand that dynamic at the time. But when we actually launched a full year later, we had, we did yet another, we replicated the announcement virality and did it again. We just ran the same playbook and it just worked again. And, and so another good moment. And then we wound up doing $30,000 in monthly recurring revenue in the first 90 days of launch. And then the conversations with VCs completely changed. It went from passive, okay, cool. We want to know you. We want to know what's going on. We'll stay in touch, but we're not going to lead to multiple competitive offers to lead. And so my advice to people when people ask like, How do you make, how do you fundraise or how do you get investors to give you money? It's you. The answer is it's to make yourself fundable. Like your job is to make yourself fundable. And to be honest, before we actually had people paying us, you know, we weren't so fundable. We weren't a known quantity and we hadn't proven that we actually, we can give a demo, but like, how do investors know that product that we were building for this post-production market? They're not experts in that market. They don't know if people are going to love that and want that. I knew that I had conviction around that, but they didn't. So by the time we launched and people were paying us money, now we had the evidence like, hey, look at this. All these people are willing to pay us money and $30K monthly recurring revenue in the first 90 days of launch is— I didn't really understand it at the time, like really good. And that was because we were selling $15 a month plans. So you can do the math. We had thousands of people that had signed up pretty quickly.

Yeah. And take us through from that launch, $30K MRR in the first 90 days to then selling to Adobe. What was the story there from scaling up, growing that, and how do you go from, all right, you got a good launch to a billion-dollar exit?

There's a lot in there from launching to selling. There's lots of different chapters and ups and downs and things like that. But we, once we launched and we raised our seed round from Accel, we won, we became anointed by Silicon Valley. We also got a great high-profile seed investor, which is, by the way, it's cool, but it's also a little dangerous. One of the dynamics that I didn't really understand at the time, but I understand a lot better now is that when you do a seed round from a very large institutional investor, they, if they don't, for whatever reason, if they don't want to invest in the next round, it signals to the rest of the market, there's something wrong here, and you shouldn't do it. Whereas if you raise money from angel investors, they only do those angel rounds. And so if they don't do a follow-on round, you have another you're, you can go at bat again and tell your story again to a different group of people. And they're not like worried that your micro angel investor is not following on to the next round because they don't do that. So there is some additional risk when you raise from a big institutional investor. Anyway, we didn't have that problem. They, we, we continued to grow, but the, the high level, the stats of the company were we, we launched in 2015, March of 2015, and then we exited to Adobe in November We're like October, November of 2021. At the time of sale, we were about 360 employees and about a $60 million run rate. And, and I'm happy to tell you about any of this stuff that went on in between, but there's, there's a lot.

Yeah. Yeah. I would love to learn about going from launch and then getting inside Studios work. I know you obviously had some connects there, but what was that like?

Not really, but not, I guess I think that what I had was the inside track that I had was I had a little boutique post-production company and I think that what I had was access to the community and the community is never like the studios. The people that are doing the work, that's not like the executives. I didn't know any executives. I didn't know any studios, but I knew the community. I knew the people that went to the trade shows and I knew the, the editors and the colorists and the VFX artists. Like I knew the community and that's what we leveraged. I'd say that's what we really leveraged most to grow. It was also a very bottoms-up company, like in terms of our go-to-market strategy, it was bottoms-up. We did not sell top-down. We didn't go out to the studios. We didn't go out to the executives or to a CIO or a CMO or whoever those people were. One, we didn't really know them, didn't have access to them. And, and I guess even in the beginning, the early years of Frame, we, I didn't even necessarily think we would ever sell to them. I, we built a self-serve product that you was inexpensive and you could sign up with a credit card. And I think it was a little bit of I think I was just a little bit naive. That's the only way I ever built, bought products. I wasn't, I never came from a big company. I was never an enterprise buyer. I didn't know what that even looked like. So I built, me and my co-founder built a product that was sold the only way I've ever bought, which was you sign up with a credit card and off you go. And so we had this very kind of bottoms up strategy and it was like we were selling to the individual. We were selling like our core person we were selling to was the button pushers doing the creative work. Editors, colorists, VFX artists, whoever. And that sort of got us into these organizations and then it would grow Classic kind of landing and expand Motion. It, within a couple of years though, within a couple of years and through kind of the guidance of our board, we had a great board. We, we did evolve into being an enterprise business and that took time to learn how to become an enterprise business. We had to I remember actually, it's funny. I remember some of the early conversations I had with Steven Sinofsky, who from Andreessen, when he was asking me about like, how do you scale this to enterprise? And, and I was so naive. I was just like, what do you mean? It's the same product for them. What are they, it's like, basically all going to be the same stuff. I'm like, I think it'll take us a couple months. We'll be ready to, you know, sell to enterprise. 6 years later, we're still building all the stuff that enterprises need. They're very demanding. They're a very demanding customer. And so it was a journey to go from being a self-serve, bottoms-up distribution strategy to being an enterprise product, both on the product side and the go-to-market side, an evolution. We had to build a sales team, had to build a bunch of features within the product that, that makes it enterprise ready. A lot of typical stuff, a lot of the stuff that enterprises need for to buy software are fairly standard across all types of software, security and compliance and things like that, but it's still hard, maybe slash annoying stuff to build. It's like not the fun stuff to build.

Yeah. Yeah. Bottoms up because you had to, I think is a really cool strategy that somebody, an outsider who doesn't have connects to a CMO, to a soft, big software buyer is something that people could really take away.

I guess you could say we had to, but I'd almost say it's like, didn't even know any better.

That's a good, that's a good way to be. Yeah. I feel like the. Negotiation with enterprise contracts, that, that must have been, that's, that sounds like a headache. That sounds long processes and stuff like that. Did you guys?

Yeah. And we just had to learn. We had never done it before. I had the, although the very first employee. So when we launched, it was just me and my co— we, when we launched, I guess we had technically by the day we launched, we had me, my co-founder, this one other engineer that we found in Germany, like an intern, and then one salesperson, actually the salesperson, the first seller that I hired. Started on day one of the launch of the company, which is weird. I don't know that typically you would do that, but I thought we're gonna need help selling this thing. And he was somebody that sold me hardware and software. He was like someone that used to sell stuff to me and I, he was my favorite seller. I had lots of people that sold me stuff and he was, I always enjoyed dealing with him. And so I went out to him a few months before we launched and tried to recruit him and hire him and he joined. I say hire him, but nobody was paid in the first, like, few months until we raised money. We gave people equity, but we didn't— we actually gave people $0 until we raised venture capital, which was only a few months. I think for the first person, this one guy in Germany probably worked for us for 6 months for no money. And then I think our first seller, Lee, worked for us for 3, maybe 4 months for no money. But even when we started paying people, it was very little.

Wow. That, I feel like that's pretty untraditional these days for somebody to take the chance on that. It's a hard thing to do.

Yeah.

I really was checking out Frame. So I use it, we use it at Product Hunt. I use it with my editor for other stuff as well. I didn't realize all the camera integrations. That's crazy. How did you guys even, that's just mind-blowing. Did you have to build out a whole special team for that in general to Yeah, we did.

I think on the technology side, it's like a relatively straightforward thing. We integrate with cameras. So when you shoot with professional, like professional-grade cinema cameras, you record on the camera and the clips show up in Frame.io all wirelessly. We call it camera to cloud. And it really is, if you think about it, it's not that complicated. It's a file transfer, right? We're transferring the file from the camera to the cloud, which we do all the time on all kinds of technology devices. That's the surface. The actually getting it in the cameras and having it work rock solid under a variety of different scenarios, circumstances, technology platforms, and then a different type of customer is using that part of the product with very different expectations than our standard customer. We did wind up building a whole special team around it, not only to build it, but then also to deploy it, in the field and support it. So yeah, but it's a program that we've been doing for a few years and we're in a bunch of cameras at this point. And it is very cool. It doesn't take a lot of imagination to think that like in the future, we're just, that's just how it's always going to work. You're just going to shoot and footage is going to appear in the place that you need it. And you're not going to have to go through all the steps of downloading it from the camera and shipping hard drives all over the country or the world.

Yeah, I like that future. Definitely seems a little easier. So speaking of future, what's your day-to-day look like right now? What's been routine for writing bangers on Twitter? As I've said, you've been doing really solid on that.

Yeah, thanks. So I'm still at Adobe where Frame was acquired by Adobe and I'm still at Adobe day-to-day. Obviously things change as you're joining a bigger team and it doesn't, my day doesn't look very looks quite different from what it was when Frame was still an independent startup. And yeah, I think that there were— on Twitter, I've been sharing more of my lessons and I think the reason I started tweeting, which was I actually didn't have any, I didn't have any day that I, or plan where I was like, I really should start tweeting. I shared one tweet that happened to do well and then I was like, oh, maybe I'll should just write more fully formed thoughts and share them. And I think my motivation for tweeting is that I think I, I felt like I've, the majority of my career, I had a lot of people around me that were, didn't really believe in a lot of the philosophies or values of how I built Frame. And I wanted to just share what my philosophies and ideas and learnings were really just for free expression. Just feels good to share that. And also it feels good to attract like-minded people. I think one of the, one of the lessons I've had is if you're hesitant to share your true ideas. Because you're scared of what people may think. I think that's not a great strategy. You really actually want to scare away the people who disagree with your philosophies. That's actually one of the objectives. The more honest and authentic you are, the more you will repel people that don't agree with them. And 99%, if I had to boil down, like, what is the number one challenge of a company? It is aligning people. Like, that's it. If you could just talk to every founder and feel like, if I could just wave a magic wand and people just did what I said, this would be so easy, right? Like, fundamentally, especially if you're doing like B2B SaaS, like the technology at this point, it's not that hard, right? Like everybody can do it. There's lots of capable teams that can do it. It's not to say that it's easy, but it's, we're not B2B SaaS is not, we're not inventing like new science. It is execution-oriented technology. And I think that most companies, the number one challenge is just alignment of people. And so if you can be very authentic with your ideas, philosophies, values, repel the people that don't agree with them and attract the people who do, you are going to have a much, much easier time building, building your company.

Yeah, that's same with building an audience, building a newsletter, whatever it is. Not being afraid to be polarizing, push out the wrong people who don't care, who don't fit. My meme marketing, meme lord stuff, like pushes out a lot of people, but it also attracts the right people. And it's just by being authentic with who I am and myself.

It'll also make you much happier if you're, if you try to, if you try to lead in an inauthentic way, it's not very fun.

And it'll be something that over time You know, yeah, speaking of authentic leaders, somebody at Adobe that I've looked up to for the last few years, Scott Belsky. Um, not sure if you get to work with Scott at all, but his Implications newsletter, make sure to like read every edition. Do you have any thoughts on Behance or just the future of any predictions of your own when it comes to that kind of stuff?

Scott, so yeah, so Scott Belsky is the, actually the executive at Adobe that led the deal for Frame.io. He was the principal on the Adobe side who initiated the deal, sponsored the deal, and was the executive shepherding the deal all the way through. So Frame joining Adobe is Scott Belsky's doing. And I'm very grateful for Scott to believe in us and bring us in. And I've gotten the opportunity to work with him over the past few years. And he's great. The right sort of what he seems outwardly is exactly how he is to work with. And, and so it's been a lot of fun and he's, yeah, I think he's especially good at, he's great at really paying attention to the market and predicting where things are, predicting where things are going. And of course, I read the Implications newsletter as well. What you're asking me, what my predictions are for, like, what do I see as future trends in the market in general?

Exactly.

Obviously the thing that is most prevalent is everything that's happening in AI, and I don't proclaim to know where things are headed with that. I think there's so many people that have opinions on it, and to me, the only honest take about where things are going with AI is we have no idea. I really think that's the only honest take. You can, from the inventors of the transformer architecture, Ilya and and Geoff Hinton have wildly different predictions than some of the other, like, godfathers of AI. Like, there is not a lot of consensus from the deepest experts. I guess this is actually getting more into the further implications of AI and AI safety. And is it going to end civilization? Are we going to hit the singularity? Like, all of those types of things. I think on that front, the only honest take is nobody has any idea. And I get a little annoyed when I see people that have such strong opinions one way or the other, rather than just saying like, how could we have— I don't know, I don't think— I just don't think there's— I have no clue. But in the short term, I actually even say that in the short term, I talked to a lot of investors, and people are really confused about how to invest in AI. Like, where does value accrue? Like, where is the— is it going to be the foundation? A lot of people think the foundational models are not going to be very monetizable. They're going to be a commodity, especially with Zuck at Meta just investing so much money and then making these models open source. So it's like, where, what, at what layer is the value going to happen? And I think we're all still figuring it out and lots of dollars are being burned along the way. I'm personally being very cautious and I haven't done a lot of AI investing just because I don't actually have a clear idea of how it's going to, where it's going. But I guess I have a couple, I have a couple maybe spicier I don't know that they're going to happen, but possibilities of the future, which is it may be that we have one more human cycle of traditional software. I think that, that software and engineering and coding is very much a task that AI is getting extremely good at extremely fast. And a lot of enterprise B2B software especially is just very much like solving business problems. And whenever we, how soon until we have generative apps where I'm a user at a business and I just type in, hey, I need an app that does X, Y, and Z, and it can just generate it. I think that it doesn't take a lot of imagination to see that will happen. Exactly what timeframe that happens, I don't know. How do traditional sort of technology companies play a role in a future where that exists? I, I don't know, but I think a lot of SaaS software might be replaced by generative software.

I could definitely see that happening. I'm already seeing there's a cool startup out of YC called Marbleism right now. Shoutouts to the founders there, but just type in builds your database, builds the front end for you, a little prototype. It's oh wow, that's crazy.

Yeah.

Very quickly. Yeah.

Yeah. And this is a category where we have tons of data, like the training to do this. Like we have, this is, this data exists and, and it's getting better at a very fast pace, which makes it hard. I think for all the entrepreneurs out there, like. Where to jump in, I think, is more challenging than ever. Like knowing where to jump in, what type of business to build, whether you're going to go like in the deep end and be a foundational AI thing as a startup, almost impossible. Like you, the amount of money you need to be a foundational model is probably at this point, like you can't do it. Maybe somebody will jump in. Although there are some counterexamples, like you look at Midjourney. Midjourney has, I don't think they've raised a single dollar. Externally, and they're obviously doing very well. So there might be some outlier examples, but it seems, seems pretty hard to jump in and say, I want to be a foundational part of the AI ecosystem. And then you say, okay, I won't be able to raise the dollars. I won't be able to spend a billion, let alone $10, $20, or $30 billion a year on compute to compete with Meta and xAI and OpenAI. So you say, okay, I'm going to build a layer on top of it. And a lot of those feel risky, like they're short-lived. When does the foundational model replace the layer that you're building on top of it? I don't know. I got— I think it's a really tough— as exciting as it is and how much there's dollars flowing out there, like in that respect, it's easy. But I think knowing what to build and what's going to have longevity over the next 3 years, I think is harder than ever.

Yeah, I, I think I had this tweet a little while ago. It's just, I, I think a lot of people are jumping into hard tech because of this, cuz they have no idea what to build in AI. And so we're seeing this like giant hard tech thing on Twitter. Everybody jumped into it and I think it's just cuz people are so confused about AI that now hard tech is hot cuz we could actually—

and I think that's awesome.

Yeah.

I mean, that's actually, that's, I've been, my, my interest has shifted over to hardware and deep tech. And me too. That feels like an area that you understand if this could exist, this will be valuable. Whereas software, just, I don't know how the software landscape's going to change, which is why I invested in one of the companies I'm most excited about that is, is Figure, the humanoid robotics company, which I invested in and, and making some other hardware and deep tech investments. The humanoid space is something that I'm really excited about. I think that's something that most of the world is just asleep on. They don't understand what's happening in humanoid robotics. It feels— if you talk to somebody, and I talk to people about it all the time, everybody's idea of humanoid robotics is some science fiction thing, or, oh, I remember that. Yeah, I've seen some like Honda robot walk around on stage, and I remember it was like $10 million per robot or something. That's like the vision that people have, but it is not what's happening right now in the real world. There is an explosion of robotics happening. In all form factors, humanoid being one of the most exciting ones, but in all form factors. And, and it's very real. These things are working and they're the big, the big unlock. There's multiple big unlocks, but the big unlock is that we now have AI models that can power them. Because even if you could build the hardware inexpensively and which before years ago, which we couldn't, but even if you could, there was no software that could run these things and have them do tasks. You can't program it's very hard to program a robot to pick up something and put it— just move objects. But now these end-to-end neural nets are training robots how to do real work. And we've already demonstrated, and whether it's Figure or Tesla or the other companies that are doing it, it's— we've demonstrated this stuff works with the technology of today, the models we're building today, the end-to-end neural nets. And so in a couple of years, I think we're going to see the proliferation of robotics. Humanoid robotics specifically, and 5, 10 years from now, probably just see them all over the place in our everyday lives.

Yeah, I have a friend who is the new head of growth, I think, at 1X or X1 Robotics, humanoid robotics company, and was explaining how the robot can look at how I do a task and then learn how you do that task. And that was the first time where it really clicked. It was like, oh, it's not just programmatic, it's AI and learning. It's not just—

yeah. And NVIDIA has— right. And the big problem is that we don't have that data. We don't have enough human data of people doing those tasks. And so NVIDIA has entered the space and they want to build the, they want to build the end-to-end neural net for humanoid robotics. And obviously Tesla's in it and Figure's in it and all these other companies that are doing it. And right now, all over the United States, there's warehouses with people doing tasks with motion capture suits on. And we're trying to bootstrap the data. There's companies that are trying to bootstrap the data and gather all this information so we can start deploying these robots into real-world scenarios. A lot of people are targeting warehousing and manufacturing as, 'cause they're fairly easy repeatable tasks. Like warehousing is like move box from this pallet to this pallet. And, and so they're being trained on those types of things. I think you've, we've all seen the videos of Tesla, like training how to fold laundry and things like that. So it's going to be, it's going to be a super, super interesting time.

Yeah, very excited for it. I think I was talking to my fiancée about it. The idea of having one of those in the house is a little weird, but when you say, okay, you could do all our cleaning for us, it's oh, sick. All right, let's go. Let's do this. If it saves that much time or if it helps, whether it's whatever, I think, I hope at least the cyberpunk in me, I hope it's not just in factories. I hope it's coming into the houses soon.

I think it will be. And I think more broadly what people— I don't see a lot of people talking about, there's a little bit of chatter about it, but you really just have to say, okay, speaking of like implications and where things are headed with AI, if we zoom out to the kind of the bird's eye view of what's going on is that if you think about all things in life, really require 3 inputs. Requires intelligence, like somebody to think about something or design something, right? There's intelligence. There's the need to transfer that intelligence to the physical world through physical labor. It's— you can't just design something in a computer, but to manifest it in the real world requires lots of physical labor. And those 2 things require energy. And so energy, intelligence, and physical labor are the inputs to literally everything. Like every good service, every kind of aspect of humanity and civilization are those three things. And we can see that those three things are on a trend that will, in the limit, probably cost close to zero. And when you have intelligence, labor, and energy that are, let's just call it near zero, we would— some— I've heard of some people start to use the term that maybe maybe in the next decade or so, we'll enter the age of abundance. And the age of abundance is when basically everything becomes free, goods and services become free. And why would they— why will they be free? Because the compute to run the intelligence is really going to be so inexpensive, it's just going to be energy as the input. And we're going to have humanoid robots that are able to do the physical labor component. And we'll live in a, in a utopia, maybe, or maybe everybody will be miserable because we won't have any We won't have to work, we won't need money. But I, it sounds wild. It sounds out there and I have no idea if that's going to happen. But again, if you just say, take the trend that's happening now and just project it forward and say, okay, our models are getting smarter. They're getting cheaper. Good. Big question is, are we going to reach that AGI moment? Is it going to really truly be as smart as a human? Is there some limiting factor or some, something different about biological intelligence? Maybe we never get there. Seems like we will. That's, I don't know, seems like we will, but who knows. Anyway, the next decade might be more different than the preceding thousand years.

Yeah, I would, tweeted this a while ago, something about if you believe AGI is coming and we are entering that, now's the time to start a podcast because people are going to have a lot more free time, man.

They're going to have a lot of time. I'll tell you, I believe it's coming, but if you believe it's, if you believed it's coming. I think that we might all be doing different. If you truly believed it was coming, then I think we'd all probably be acting differently. And I cannot get myself to act differently. I believe it's coming, but I'm not actually acting any differently in my life. I'm not really doing different things in my life. Yes, maybe from an investment perspective, I'm looking at like where the sort of next 5-year opportunity is, but I think it's really hard, even with a lot of conviction about where all, where this is trending. To like truly internalize that and imagine that just that future that I just described, it's almost impossible for us to really grok.

Yeah. I think bullish on relationships, content, community as well, I think are all super important things.

Probably. Yeah. I think people will probably tend to value human stuff.

More and more. Yeah. I've already seen that with friends who are leaning towards the digital minimalism life, which I've explored at times, and now I'm like digital maximalism at this point with my Twitter addiction. But yeah.

Yeah, me too.

I think, yeah, so it goes. Yeah. Are there any other industries you're interested in, like super investing in besides humanoid robots?

I'd say robotics is the one I've been paying the most attention to and trying to learn the most about and the intersection of robotics and AI, and then AI in general. I think there's a lot of interesting stuff happening in energy, and I'm not, I'm not the expert there, so I haven't gone, I haven't gone deep, but I've, you know, you look at deals with these companies that are gonna, they're like literally inventing new physics, and you're like, okay, I, I'm not a physicist, and they're really hard. It's, they're, it's, I've stayed away from those deals, but there's a lot of stuff interesting that's happening at the very frontier of science. And I think figuring out the right way to invest in those companies is something I'm still, I'm not there yet. Yeah.

On the Sam Altman prediction of a one-person, $1 billion company, do you think that's possible?

Yeah, absolutely. I think, I absolutely think it's possible and I think it'll be short-lived because I think those are going to be, they'll be, that might happen. Like once or twice, but the models are that, that basically what that means is there's going to be one person that's just a little bit better at pulling the models together so they can leverage all the upside. But then I think the models would get good enough where everyday people will be able to utilize them in the same ways that like a one-man or one-woman kind of engineer would be able to do it. You think it's okay, you're going to get there because in the next couple of years, there's going to be one person that's going to be leveraging design models and things for marketing and certainly things for engineering, and they're gonna be just outta reach for your everyday person to bring together. And big teams are not gonna move quickly enough to adopt them. And there's gonna be this sort of one person that's, oh, I get this, I can do this, and I'm gonna find the right vertical of a thing people need. And look, I think you see this a li— this is not a one, this is not a one-person operation, but Midjourney is very small. They're a very small team. And I, as far as I know publicly, I think they're like doing like $100 million a year. And certainly that would be valued well over $1 billion. They're not taking any external capital in. And these numbers may be wrong. I just want to say this is what I've, you know, heard out there. But absolutely there'll be people that build a, it doesn't even have to be a one person. If it's a five person company, that's what's the difference? Like five person, one person, it's going to be a, it's going to be a couple people. Maybe eventually it's a one person. But the point is, I don't think that opportunity will last for very long.

Interesting. Interesting. And closing things up, one more question is I have a Vision Pro sitting on my shelf over there. I haven't used it. You got one too? When's the last time you used yours?

Yeah, to be honest, it's been a, it's been about a month. I don't think I've put it on. I was really heavy utilizing it in the first, when people were saying like, ah, I used it once and I put it down. I'm like, what are you talking about? I'm using it every day. And I was, I used it a lot. For the first maybe month, maybe two, but it's been sitting on my desk now for a while. And I think I'm still very bullish on, when you put, here's the thing, when you put that headset on, it is so obvious, it is blindingly obvious that this is how you want to have this interaction style with computers. And it feels very power, you feel very powerful in it. You feel like you're in The Matrix. It's awesome. And certainly the media consumption experience is awesome, but you just don't want to put that thing on your head. And so that thing's got to get lighter. And I'm, I think it's actually a really good time for companies. I think that's an interesting area to, for startups to be investing in that because right now nobody's paying attention to it. You can like, you want to be, imagine if you were the people that were like before when everyone still thought AI was nothing and you spent those 3 years just before that getting ready. I think that AR and VR is there now, and people would have said that there was a moment, there was a moment in time 10 years ago when that people were saying, oh, it's going to be the next big thing. But it is different when Apple enters a market. Apple has commitment to the hardware. I think they're very committed to the amount of money they've spent on this. It would be extremely surprising if they didn't go through several cycles to try to get this to be somewhere good. And maybe it takes multiple forms. They're going to have a fully immersive one. Maybe they're going to do glasses. That are, I think the pass-through is what we really need. Like the real light pass-through is what we really need for them to be even better than they are. But anyway, I'm still bullish on that as a future computing platform. I think the hardware that exists right now is not going to get us there.

Yeah, I agree. The first time I put it on was just, oh, blown away. And this is it. This is the future. We are here. And then you start playing with and all the bugs and like the little things. And just the friction to put that heavy thing on. I don't even wanna put these things on heavy anymore.

Exactly.

Once you've had AirPods for years, why am I putting these giant Bose on at this point?

Yeah. I don't know why people do that. I'm AirPods all, all the way.

Yeah. I, I gotta go find mine. They're somewhere in my, they're too easy to lose. They are. That's true. Yeah, dude, this was awesome. I'm gonna, I'm gonna stop it. Really enjoyed this conversation.

Cool, man. Thank you.