--:-- Host (Erik Torenberg): We wrote our first lines of code 10 weeks ago. We're like earlier than the latest YC batch of companies, yet we're generating probably more revenue than every single one of them.
--:-- Host (Erik Torenberg): It's been so hard to pierce through the noise of everything in AI, especially for the consumer facing side. To do that consistently is actually way harder, near impossible.
--:-- Guest (Roy Lee): I heard someone call it "Rizz marketing," which is a compliment. They're like, "I hate this Rizz marketing."
--:-- Guest (Bryan Kim): AI is so magical. We built the digital god, locked it in a chatbot.
--:-- Guest (Roy Lee): Six months ago, I was some random college kid in a dorm, and now I feel like I'm at the center of the tech universe.
--:-- Host (Erik Torenberg): What happens when a founder treats virality not as a tactic, but as the product? Roy Lee, co-founder and CEO of Cluely, is either a generational entrepreneur or a walking internet experiment. Maybe both.
--:-- In just a few months, he's built one of the most talked about startups in tech — not with a massive fundraise or a polished product suite, but through relentless short-form content, polarizing stunts, and a translucent AI overlay that feels more like performance art than interface.
--:-- Joining me today is Roy, along with Bryan Kim, who led A16Z's investment in Cluely. We talk about Roy's approach to building in public, what Gen Z understands about attention that tech still doesn't, and why momentum might be the new moat in consumer AI.
--:-- This is a conversation about distribution as design, founder-market fit at internet speed, and what happens when you stop trying to be professional and start trying to win. Let's get into it.
--:-- [Disclaimer: The content here is for informational purposes only, should not be taken as legal, business, tax, or investment advice, or be used to evaluate any investment or security, and is not directed at any investors or potential investors in any A16Z fund. Please note that A16Z and its affiliates may also maintain investments in the companies discussed in this podcast. For more details, including a link to our investments, please see A16Z.com forward slash disclosures.]
--:-- Host (Erik Torenberg): Roy, man, the myth, the legend, the man of the moment. Clearly the current thing.
--:-- Guest (Roy Lee): Yes.
--:-- Host (Erik Torenberg): How does it feel? The announcement the other day — a lot of love, a little bit of hate. How do you react to this?
--:-- Guest (Roy Lee): It's pretty crazy. Literally six months ago, I was some random college kid in a dorm, and now I feel like I'm at the center of the tech universe. The more astonishing thing is how correct my assumptions on virality have been.
--:-- It's growing increasingly clear that people on X and LinkedIn are behind. There's such an extremely small intersection of people who understand how developed the algorithm is on Instagram and TikTok, and people on tech Twitter and LinkedIn. That intersection is so small that it's inevitable that my predictions will be right. It's been crazy to see that play out in real time.
--:-- Host (Erik Torenberg): Elon's reaching out. Meta's offering a billion-dollar acquisition offer. Bryan, we said not to do it, of course.
--:-- Guest (Bryan Kim): Of course. To the moon.
--:-- Host (Erik Torenberg): Bryan, what is your reaction to seeing this play out and having led the investment?
--:-- Guest (Bryan Kim): Well, first, he's a bro of the moment. I love the little snippets of you calling Molly "bro" like 20 different times. Amazing.
--:-- Look, Roy, this is fantastic. It's so interesting. I remember right before the announcement, you had already sent me the video 24 hours ago — "bro, this is what we're going to show." And I was like, okay, this is great. I thought it was going to do well. I didn't think it would do *that* well. And I didn't think it would create that much controversy.
--:-- But I also underestimated how positively people would view this. There's a lot of meta-analysis on you out there, which is really cool. People are going two to three layers deep. I actually read something that linked a meta-analysis of A16Z with Cluely and how there's this fungibility of thoughts and things like that. It's mind-blowing. I have no idea what the fuck they're talking about. So it's just been incredible.
--:-- Host (Erik Torenberg): You put people doing literary criticism.
--:-- Guest (Bryan Kim): No, I'm not kidding. There's literally a guy who wrote about the fungibility of thoughts with A16Z and Cluely.
--:-- Guest (Roy Lee): That's funny. People always over-read into stuff. They're like, "Oh, this has this master plan. A16Z is trying to do this whole thing."
--:-- Guest (Bryan Kim): That's what I'm saying.
--:-- Guest (Roy Lee): Don't tell them. We are playing 4D chess here. There are ten threads analyzing the left-hand handshake. It's like, guys, come on.
--:-- Some of the haters... I tweeted something. I was talking to Bryan. I tweeted, "How you feel about Cluely is how you feel about yourself." But I deleted it because I didn't want to trigger people on Twitter. I'll talk shit on a podcast. Maybe I'll get clipped to Twitter, but I don't want to inflame people.
--:-- Host (Erik Torenberg): It's fascinating to see. I want to go back to the beginning because this isn't a story that has been told a lot. We were talking about the Amazon interview, but maybe let's go back even further. Talk about the threads or the through lines from your childhood that can help make sense of who you are and this moment.
--:-- Guest (Roy Lee): From birth, the most character-defining feature of me has been attention-grabbing and provocative. This played out in elementary, middle, and high school. I always had a camp of people that loved me and a camp of people that hated me. I was always the boldest. I would say the craziest shit. Everything that was on my mind, no filter, I would say it. This ended up with a lot of people liking me and a lot of people really disliking me.
--:-- Things culminated senior year of high school. I had a well in school. I got accepted to Harvard early. Then later that year, I was always doing crazy shit. I snuck out of a school field trip past curfew. The police had to come and escort me back because we were out at 2 a.m. We were all like 16, 17, 18. I got a suspension for that. That was when the camp of Roy haters took the storm and reported everything everywhere. It ended up getting rescinded from Harvard.
--:-- That started my journey of wanting to swing big at building companies. That's when my life took such a crazy turn. For context, my parents literally run a college admissions consulting company. We teach kids how to get into Harvard. The youngest son of the company gets fucking rescinded from Harvard. It's the worst thing ever.
--:-- We decided to keep it quiet. You still have the same test scores and application. Maybe next year you'll get into a different school. So I spent an entire year at home. I underestimated how mentally tormenting that would be. I'm probably the most extroverted person you might have ever met. I cannot stay maybe eight hours without talking to someone. To spend a year alone made me think, "Man, my life is so crazy. I might as well just double down on every single crazy belief thought I have and just live the most interesting life ever." That was the moment where I decided I'm all in on building companies. There's no way I can do anything else.
--:-- Host (Erik Torenberg): I was wondering if you were going to have a moment of, "Oh, this has set me back in some ways. Maybe I should reform or tame it down." But you took the opposite — "No, this is who I am, and it's going to work."
--:-- Guest (Roy Lee): You just sit in a room by yourself for 12 months and all of a sudden your craziest thoughts become logical. There's no one else. The echo chamber is you in your brain, and it amplifies everything. I think that's the reason why I am the person that I am today and willing to make the bets that I am today.
--:-- Later, I go to community college in California at the behest of my parents. California is a middle ground. I have the chance to build a company in community college, and I have the chance to get the education that my Asian parents always dreamed of. So I do that. Then I get into Columbia, and I have to go to Columbia for at least a semester to appease my parents. I go to Columbia and the first thing I'm thinking is, "Can I find a co-founder and a wife?" Those are the only two things I'm looking for in college. Still looking for the wife.
--:-- On pretty much the first day, I met Neil, my co-founder. We started hacking on a bunch of things. The one thing that worked was the earliest version of Cluely.
--:-- Host (Erik Torenberg): Were your parents ever trying to reform you or mutter at you? Or were they kind of like, "Hey, Roy's going to be Roy"?
--:-- Guest (Roy Lee): Almost every single day of my life until I got into Harvard. Then they calmed down and were like, "Wow, this kid really made it." When I got kicked from Harvard, they were on my ass a lot until I got into Columbia again. They were like, "Wow, after all this, he gets back into the Ivies. I guess I really can't trust him. And his unorthodox swings will be home runs."
--:-- Host (Erik Torenberg): Is that where they're at now?
--:-- Guest (Roy Lee): It's crazy how lax they've gotten since I got back into Columbia. Now they're okay with anything. When I told my mom, "Hey, I'm dropping out to do this," she was like, "Oh, okay. I expected it. Why didn't you drop out sooner? Why did it take a semester and a half?" I was convincing my co-founder to drop out with me. She was like, "Man, took it long enough." So they're totally on board with all the crazy shit that I do.
--:-- Host (Erik Torenberg): One of the things we were talking about in the context of going viral — I heard you say that Twitter is two years behind Instagram, or behind the other platforms. Talk a little bit about when your provocativeness translated over to Twitter or the digital mediums. How did that strategy evolve? And let's talk about the difference in the platforms.
--:-- Guest (Roy Lee): This goes way back. Many years ago, when YouTube first came out as a platform, that was the turning point of everything. It democratized content. You weren't paying for commercials. Visibility and publicity in content was not gated by the amount of money you're willing to spend on ads or TV space. It was gated by the quality of content.
--:-- Five years ago, when TikTok came out and short-form algorithms really started taking over, that shifted the frame once again. Now it's not about how much good content you make. It's literally just about how much content you *can* make. There is simply not enough good content out there for the average person to consume, which is why you see the same brain rot reels over and over again. You see the same Minecraft parkour video over and over just because there's literally not enough content for the average consumer to consume.
--:-- People have not caught on to a few things. First, you just need to make more content that a consumer will like. Most people don't know how to make viral content — content that any person can watch, consume, and is digestible. Everyone on X and LinkedIn is trying to be the most intellectual, thoughtful person. They generate some slop that maybe 200 people in the world can actually understand. Of course, they want to seem like the most interesting, thoughtful, intelligent person, but this lacks viral sense. There's not enough viral content to go around.
--:-- The second thing is that the algorithms really promote the most controversial things. People on X and LinkedIn — there's not enough controversial things to be rewarded for. So when I come out swinging out of the gate — I've been on Instagram and TikTok for the past 10 years of my life, and I understand what level of controversialness you need. I take the slightest foot into controversialness and all of a sudden X and LinkedIn explode because the algorithm inherently highly rewards this stuff. It gets shoved into everyone's feed and they don't understand. I'm literally applying the same principles of controversialness from IG and TikTok onto X and LinkedIn, and they're just so not ready for it that it feels like the craziest thing ever.
--:-- I guarantee my videos do not go as viral on Instagram and TikTok. The sole reason is because on those platforms, they are not controversial enough. On those platforms, there are literally people committing felonies in public, or at least insinuating they're committing felonies. Even then it'll be like, "Oh, good try, bro. It's not interesting enough." X and LinkedIn have just not caught on. There's not enough creators out there who are willing to press the controversial button.
--:-- Host (Erik Torenberg): Mark [Andreessen] says this sometimes — the supply chain of the meme. It's the strangest concept, but the meme actually travels from Reddit, then to X, then to Instagram, then LinkedIn, then CNBC. There's a train that it goes on where certain people are just early or late. But when you flip it with virality and controversialness, maybe that flips a little bit.
--:-- Guest (Roy Lee): Well, sometimes it starts with 4chan or something — the most controversial.
--:-- Host (Erik Torenberg): That's what I mean.
--:-- Guest (Roy Lee): I didn't even want to say it. It's 4chan, Reddit, then Twitter, then Instagram, then LinkedIn. But for you, Instagram actually comes before Twitter in terms of the raunchiness or craziness of it.
--:-- Guest (Roy Lee): I just feel like the average person who's in X comments hating about how controversial this is — they spend one hour looking through the timeline that is my Instagram feed. Their brains would melt and explode. They would not be able to comprehend how people are digesting this at scale.
--:-- Host (Erik Torenberg): It's funny because ever since Elon took over X, some people started complaining, "Oh, this stuff has gotten too controversial or too much dark stuff or too much negative content." It turns out it's just the beginning.
--:-- Guest (Roy Lee): Literally, this is what the future of content is going to be. You're not going to get more millennial founders. You're only going to get more Gen Z founders. I guarantee you their backgrounds and content are the exact same as mine. They know exactly what I'm doing. I'm sort of like the canary that's leading the way for this. I guarantee you there's more of this coming. It's inevitable. Just embrace it.
--:-- Host (Erik Torenberg): When did you realize that this was the way you were going to build a company? How did you intuit that distribution is a scarcity, that distribution is what matters? There are a lot of creators out there, but they're not combining it with a tech company. How and when did you put this together?
--:-- Guest (Roy Lee): There was a certain point where I kept going viral that I realized I know something that X and LinkedIn people don't know yet. It's sort of a mastery of the algorithm.
--:-- Everything started with the Interview Coder situation. Interview Coder was the earliest prototype of Cluely. It was a tool to let you cheat on technical interviews. I used it to cheat my way through an Amazon interview. I made it super public. I posted it everywhere. It ended up getting me blacklisted from big tech and kicked out of school. That situation was inherently viral. When's the last time someone got kicked out of an Ivy League and raised $5 million? This has probably never happened in the history of humanity. That situation was inherently viral. At that time, I had no idea that this was a repeatable thing that I could do.
--:-- But then the launch video happened. I had my intuitions about the virality of the launch video. I kept scrolling on Twitter and I was wondering, "Man, why is nobody doing what Avi Schiffman with Friend.com showed the world you could do a year ago? Why has nobody done this yet?" And it worked. Then I did the 50 interns thing and it worked. I kept doing viral video after viral video. At a certain point, I realized, "Holy shit, people on X and LinkedIn have not caught on yet. This is the massive alpha that I'm trying to capture here. They have not caught on to what it means to master the short-form algorithm." As a result, I am able to dominate the timeline for not just the past week, but the past few months, just because people on X and LinkedIn have not caught on. For some reason, they still refuse to catch on.
--:-- To put a final point on the 50 interns: At your company, you either have engineers or you have creators. There are only two roles. You are either a world-class engineer who's building the product, or you're a world-class influencer. Every single full-time person has over 100,000 followers on some social media platform. It is the only way to prove that you actually have mastery over virality and you understand what it takes. I think if any company in the world has a marketing team and the head of marketing does not have at least 100,000 followers, you need to replace them. The game has changed.
--:-- Host (Erik Torenberg): Do you think this is a strategy that other companies should also be employing — whether it's intern-based or just having an army of creators and deploying them towards their end?
--:-- Guest (Roy Lee): I'll go a bit deeper on the interns. Cluely made a pretty viral video announcing that we were hiring 50 interns and you'd be in here making content all day. That's essentially what we do. We have over 60 contractors. These contractors get paid per video and they just sit in front of a camera and make TikTok and Instagram videos about Cluely. This is what marketing looks like. This job did not exist five years ago. How do you explain the job when you sit in front of a camera and make five to ten second videos that seemingly make no sense to anybody, but just consistently generate millions of views? That's not a job that makes sense to people, but that is our internship. That's what a modern day marketing internship looks like.
--:-- We pay very little money for the amount of views that we get. Different companies are paying literally millions of dollars for Super Bowl ads when you can get the same quality and quantity of views for $20,000.
--:-- Host (Erik Torenberg): And you see it converting?
--:-- Guest (Roy Lee): Yes. Our only converting videos are the ones we have on IG and TikTok.
--:-- Host (Erik Torenberg): Bryan, why don't you share your story of how you got excited about Cluely, how you and Roy met and built this relationship, and how this partnership formed?
--:-- Guest (Bryan Kim): I had a good contact in New York who travels in the young crack folks area. Her name is Ali Deboe. One of the lists she sent had Roy in it. I read what they were working on and it reminded me of a scout thing or something on the edge of reality. I thought, "Oh, this is interesting. I want to talk to him." So I reached out. Roy, if you remember, I just reached out: "Hey, I heard your name. We should talk." You were like, "Oh, bro, let's talk." Then a day later, you wrote back: "Am I actually? You're multi-stage? I don't want to talk to you. My advisor says don't want to talk to you. Go away."
--:-- What do I do? I could have just stood down, but I said, "Okay, fine. I promise you will not talk about fundraising. Let's just talk. I want to meet you. I want to build a relationship." Thankfully, you agreed. We got on a quick call. We chatted a little bit. You had your origin story. I'm like, "Oh my God, it's amazing. It's so cool. I'm glad he's doing this." Sadly, he's not accepting money, but that's okay.
--:-- Then I tracked you. I tracked your Twitter. I tracked what you were doing. The 50 interns. You moved to San Francisco. I had it in my mind: "Okay, next time an opportunity strikes, I'm just going to show up." I somehow got your phone number or some such thing and texted you: "Yo, I'm at your office. Can I hang?" So I come and you say, "Yeah, that's great. Come up."
--:-- What I actually think was really cool — there are a couple of steps. Step one was there was an engineer who randomly found you on Twitter or Instagram and had just come up. You did not know him. He did not know you. He just came to your office to say hi. One of your friends, I think Nicholas, was just there hanging out. The quality of the people, the fact that random engineers were knocking on the door to come talk to you randomly — I thought, "Oh, there's something really strange and special happening here." All of your team members sitting in the thing, doing things, creating content, you and Neil working on the product. It was like, "Oh, there's something very special." I was thinking, "Is this something that we should back?"
--:-- Then I think one more video was made or something. Next time I visited, I came with some stuff. You were eating steak. Then you flashed some metric or something where I realized that you're converting this awareness and eyeballs into money and dollars. You dropped some numbers: "Oh yeah, we're doing this much revenue." And guess what? Some enterprise customer wants to talk to us. I don't know why, blah, blah, blah. That's when I realized, "Oh, he is actually able to convert this awareness and distribution into real dollars." I don't know many people who know how to do that.
--:-- During that time, I was already writing this thing called "Momentum as a Moat" because it's been so hard to pierce through the noise of everything in AI, especially in the consumer facing space. To do that consistently is actually way harder, near impossible. I had the theory that companies who know how to do that, companies who know how to build at that speed, are going to be the winners. I felt like I had found a person who was doing that.
--:-- We moved very quickly. I told you, "Look, just hit download on the Stripe data. Send it to us. I won't ask any more questions. Then we'll have a chat." That hopefully is what we delivered. We quickly scrambled to do a very fun small person chat with you and some of the partners where you called some of them old, bald, and boring. We were excited to do the deal. I was at an LP summit running around in Las Vegas, trying to call you to get to terms. That's how it all worked out. After a while, I brought you five or six pounds of steak as an excitement and a deal.
--:-- Host (Erik Torenberg): I want to double click on the "momentum as a moat" piece, Bryan, because you have an interesting background. You've been doing the Timber for a while. You worked at Snap beforehand, among other places. I remember Ben Thompson had this post about Snap where he said that Snap has a gingerbread strategy — basically if they invent stuff, Meta is going to copy it, so they just have to keep inventing stuff. You've also backed a number of these. You've had a front row view to the fact that network effects aren't as necessarily defensible as they once were. Companies need to keep innovating, keep pushing. Share more about how this theory of momentum, especially as it moves to AI companies, was born and what it really means for defensibility.
--:-- Guest (Bryan Kim): This might be somewhat interesting to you, Roy, and folks. I did not have that view before. I did not think the gingerbread strategy necessarily worked, nor that momentum was a moat. I truly believed in these handcrafted artisan products that really get to the core of why people want to use it. I still somewhat believe the core of it. But these artisan products where it takes a while to build and it's very nuanced — I believed that led to high retention.
--:-- The thing I look at the most always was: Is this product highly retained? Does it have a network effect on the traditional sort of theories of moat? What I realized — and this was true to some extent in the era of mobile — mobile is a two-decade-old platform. A lot of things have been tried. Finding something that people came back to again and again was the most important thing in my mind.
--:-- Then AI hit. I still had that framework where I'm going to look for things that are highly retentive and repeated again and again. Guess what? Things change too fast. The underlying model changes every day or every week. If you craft this thing and OpenAI builds their new model to include that part in their new product, you're done. You're gone. So it couldn't become about this highly thoughtful, slow build product. It needed to be something where founders knew how to move extremely quickly. That included product, that included distribution.
--:-- Because these things are so magical — AI is so magical. We built the digital god, locked it in a chatbot. Because it's so magical right now, anything goes. People will give it a chance. What's really important is to try to build the plane as it's falling down the cliff. People who enjoy the thrill of the plane going down and are actually excited about building as it goes down — I think those are the winners of the next day.
--:-- When I think about folks like Roy, it's the type of founder archetype who gets value, who's excited and leading the charge in terms of that speed — whether it's marketing, distribution, or product build. Usually all of that needs to come together to build an extremely durable, long-dated product. That, to me, eventually will turn into a product that needs to be retained, that needs to be used every day. But we're still in this early stage of AI where I think momentum is the most important thing.
--:-- Host (Erik Torenberg): Going back to Roy, I'd love to hear how you think about it. We talked about it a little bit over chat where you think about stage one, stage two, stage three of building Cluely and the distribution advantage that you have. You keep talking about how maybe X and LinkedIn people don't get it right now, but that gap may narrow over time. How do you think about the next stage?
--:-- Guest (Roy Lee): [Response continues in next segment]
--:-- Host (Erik Torenberg): I want to get to Roy's product strategy, but first I just want to add on some points. It was interesting. Paul Graham, when he started Y Combinator, identified that technical founders were undervalued. They were being underappreciated. People thought you needed to have an MBA. He realized it's easier to teach technical founders business than it is to teach business people how to build great products or how to code.
--:-- Then what happens over the next 15 years? It becomes way easier to build these things — AWS, low-code, AI, et cetera. Distribution becomes the scarcity. There's so much software. It's such a flooded ecosystem. As you quoted in your piece, Andrew Chen's piece about how all the marketing channels suck — they've all been rung out dry. Distribution is now the scarcity.
--:-- In the same way you had the technical co-founder, now there's almost like an audience co-founder. How do you really break out? We've seen creators start to build businesses — Mr. Beast with his Feastables, Kylie Jenner with her lip gloss. Commerce. But no one's really done it with software. No one has put the two and two together. Justin Bieber tried to do a social network called Shots, or John Shahidi using Bieber. But no big creator has really built a huge software company — consumer or enterprise. They built commerce or physical goods. I always thought, "Hey, why doesn't Mr. Beast launch a Square competitor?"
--:-- Host (Erik Torenberg): Like he's got all these eyeballs. I'm sure he's got to get better margins than Feastables. And he has games and he's a friend of the firm and he's done phenomenally well. But what I like is that Roy is putting both of those together. So maybe you can walk through a little bit. You talked about your distribution strategy. Why don't you talk about how the product strategy has evolved from the beginning, and then we get to the sequencing.
--:-- Guest (Roy Lee): Yeah, I think something that a lot of people miss is that the first line of Cluely code was written like 10 weeks ago. So this is really new. This started with Interview Coder, which is a product that I coded up over a weekend in my dorm. It was a tool to let you cheat on interviews. After we got about 250 million impressions on the whole scenario, we realized, wow, we just got so many eyeballs on this thing. Maybe if we can do this again, but with a more general use product and a similar UX—because we think we're really onto something with the UX here—then maybe we can make a lot more money.
--:-- Guest (Roy Lee): That's what we did with Cluely. We launched it as "Interview Coder for everything." Cheat on everything. Let's just see what happens. The usage data will tell us what people are using it for. That's exactly what's happening. Now we have this general-purpose cheating tool, which in reality is just an invisible AI overlay. It's a new user experience for AI. We pushed it out to a bunch of people and saw what happened. As a result, we're past a billion views overall on Cluely. We're probably the most viral startup in the world. We have all this usage data that literally tells us where this is most sticky and where the product direction needs to go.
--:-- Guest (Roy Lee): I think that's the core advantage of distribution. You do not have to worry about market fit because your users will tell you where the direction of market fit is headed. Their usage data will literally give you the information you need. If you don't have usage data, you're shooting blind. Every person who has built a company knows you can't really know what direction you're going. You have to talk to your users. But if your distribution is strong enough, you don't need to talk to your users. You just need to look at their data. You just need to look at the aggregate number.
--:-- Guest (Bryan Kim): You're also redefining what a minimum viable product is to some degree. Other companies will spend many months building this thing and then seeing how people use it. But for you, if you can draft the right content, you can test out the idea in a much quicker way to see if it's really resonating.
--:-- Guest (Roy Lee): Exactly. When we launched the video, we barely had a functioning product. The day before is when we finished our final test and we were like, "OK, we think this works. Now let's launch the video as soon as possible." We launched the video and all of a sudden tens of thousands—we just said, "Hey, let's throw sales calls in the videos to see if people use it for sales calls," because that seems like a pretty lucrative space. All of a sudden, we have over a million dollars of enterprise revenue coming in for people using it for sales calls. You can shoot in the dark on distribution a lot quicker and more accurately than you can shoot in the dark on product. You don't need a million product integrations. It's so much quicker.
--:-- Guest (Roy Lee): What's even better is that the iteration loop is much faster too, because the algorithm will literally tell you via a number—number of views, shares, whatever—how well your strategy is going to work. It's much easier to test. Does this have viral fit, rather than does this have market fit?
--:-- Host (Erik Torenberg): Roy, does that mean you let the audience guide where the product goes? Is that how you think?
--:-- Guest (Roy Lee): Yeah, exactly.
--:-- Guest (Bryan Kim): Let's talk about the form factor. One of the things we internally discussed is that Roy is probably top 1%—now I revise it to top 0.01% in the world—in terms of knowing how to distribute. The Venn diagram of that and people who had the instinct to build a semi-translucent overlay... It sounds so simple. The disappearing picture sounds so simple. It's not technically hard. A half-translucent overlay sounds simple. It's not technically hard. But that overlap to me was what gave me so much excitement around what you're building.
--:-- Guest (Bryan Kim): I actually have it on right now. Eric, did you go to University of Michigan?
--:-- Host (Erik Torenberg): Yeah, I did.
--:-- Guest (Bryan Kim): See, I didn't know that. But I have Cluely open. I'm like, "You went to U of M. Got that. Oh, you did philosophy, policy, and economics, I think." Great. We can bond over that. This is an incredible tool.
--:-- Host (Erik Torenberg): That's amazing. Talk more about where you see the product going, or particularly how you think about it. Anyone building AI tools asks themselves the question: how is this defensible from one of the major players? Will OpenAI or someone else just build this feature? How do you think about making your product truly defensible, especially from people who, because of their reach and size—OpenAI has distribution too, right? How do you think about this?
--:-- Guest (Roy Lee): I guess we're first to move in a pretty novel UX. I think everyone's going to inevitably get to translucent overlay. This is how integrated AI should feel. Apple shows everyone that liquid glass—the translucent overlay—will be the form factor of AI in the future. Right now, it's just a land grab. If the question is about distribution, I think there's actually a pretty strong case for us to make that we will end up distributing better than OpenAI. Enough that you could probably bet on us at a 30x discount. I'm actually not worried about distribution.
--:-- Guest (Roy Lee): I think the quality of the product—it's quite simple. I really feel like this is just a land grab right now to see who can convince as many consumers and enterprises first that they are the guy who deserves to win the translucent overlay. Right now, we're making so much noise. With the translucent overlay, why would it not be us?
--:-- Host (Erik Torenberg): When did you figure out the translucent overlay?
--:-- Guest (Roy Lee): I was in my dorm with Neil. We literally spent every day thinking about how can we make InterviewCoder more invisible to interviews. We played around with probably 20 to 30 versions of InterviewCoder in the past that we thought didn't work. It essentially feeds you a code answer, an answer to a coding problem. You need to overlay that on top of your code. We were like, "Man, I really need this integrated into my code. I need to see what I'm doing as well as see the answer that AI is giving me." Eventually we landed on translucency. It was a magical moment. This is what the product needed. Very soon we realized, why are we only thinking about coding interviews and software engineering coding interviews? This is such a small market. This is true for everything. AI should not feel like a separate window. It should be integrated seamlessly. That looks like translucency.
--:-- Guest (Bryan Kim): Brian, you said you were inspired by the Dragon Ball Z scouter as an example. Roy, I'd love to chat about the staged approach of how you're thinking about it. Right now we're distribution first, and then we'll build the product as we go. Stage two, here's how we do it with a bunch of engineering prowess and product development. We chatted a little bit about that on text. To me, that was like, "Oh, okay, you'll figure out product as you go."
--:-- Guest (Roy Lee): Yeah, exactly. Right now the internet is up in storms saying, "Hey, where's the product? Where's the product?" The two things are: we are literally working day and night to build out the product that we have in our head that the users are telling us they want. But also, every video I make that's not directly about the product drums up so much more hype for the eventual product launch video. I guarantee this will be viral. I guarantee it will be more viral than if we just launched earlier.
--:-- Guest (Roy Lee): I think there's truth in the statement "launch early, ship fast, launch before you're ready." But for some reason, when we're doing that at scale, it feels like everyone is saying, "Oh no, you launched too early." I'm like, "What are you talking about? This is the playbook." We wrote our first lines of code 10 weeks ago. We're earlier than the latest YC batch of companies, yet we're generating probably more revenue than every single one of them. The product is literally two and a half months since being built. In my perspective, we are pre-launch.
--:-- Guest (Roy Lee): The huge benefit of being massively distributing pre-launch is that you will know what product to build with as much certainty as you could possibly get. If you can distribute and hype it up to an audience of literally millions of people—"Hey, this product's coming out, this product's coming out"—and we're screaming to the world, "AI overlay, AI that sees your screen, hears your audio," the second we make that, why would you pick anyone else's product? We've been screaming it from the top of our lungs since day one, before day one even. Right now, that's the stage: distribution. Get it into everyone's mind. What is Cluely? Cluely is the invisible AI that sees your screen and hears your audio. Everyone knows this. As soon as we launch it, who else will they download?
--:-- Host (Erik Torenberg): One thing I think is fascinating about what you're doing—I'll compare it to our friends at TBPN, who are kind of rebranding. They call themselves "corporate driven media" because everyone was like, "We're independent media." And they're like, "No, no, we're corporate backed. We're more honest that way." They're leaning into all the bits about ramp and "say 5%." There's this kind of humor to it. I think similarly, you're leaning into the controversy, or being controversial as a strategy. Some people think, "Oh, that's fake or forced or whatever." But at the same time, you're super authentic in the way you're doing it. People feel like they know you. Sometimes even if there's a character, an exaggeration, that also feels authentic in some way. It's a unique style.
--:-- Guest (Roy Lee): I think this reflects a growing shift in society. For the past few decades, there's been such a sharp drop-off in professionalism. I really think it's because content creation has been democratized. The first ever YouTube creators were just people making funny videos in their dorms. This was the most authentic. People crave this. Nobody wants to see another ad or another corporate newspaper, bullshit like that. They want to see some real person doing real things. The democratization of content creation has allowed authentic people who make content to be seen by millions. For some reason, no company gets this. You'll have 100x followers and your post will be about "Introducing blah, blah, blah." It's the most corporate bullshit ever. You're not going to get any views. Nobody wants to see this.
--:-- Guest (Roy Lee): Would you rather have a world where everyone was extremely transparent, where even the founders were honest, and everything you see is just someone's authentic life? Or would you rather see a bunch of corporate bullshit everywhere? What we envision for the end state distribution of Cluely is that it does not feel like an ad at all. This is just the story of someone's life that you want to see. It is true and authentic. I'm probably more transparent about everything than 99% of companies in the Valley are.
--:-- Guest (Bryan Kim): What you're saying—some people get so mad. I heard someone call it "Rizz marketing," which is a compliment. But they're like, "I hate this Rizz marketing. It's got to be product first. Get back to fundamentals." But first, this is its own fundamentals of how the world works today. Second, of course they can't destroy you. Even the people who want to destroy you—the joke in my head is they could try to kill you, but they can't kill the idea. The way you're building companies, there's something about it that is going to have an impact on the next generation. You're just starting the company journey.
--:-- Guest (Bryan Kim): When you say viral marketing, that's interesting. One leads to another. I actually think what you're doing is anti-fragile marketing. You're so controversial that someone says, "I'm going to cut off your head," and three sprouts grow up because one is mad at you, one is really happy about you, one is neutral. Every time someone comes at you and comes at the idea of Cluely, the more aura points it gets. It's aura farming. Any lessons from the most controversial stuff you've done? Is it always triple down? How do you think about where the line is?
--:-- Guest (Roy Lee): The few lessons have been: never punch down. Never even remotely close to punch down. I think people reward—and the algorithm does reward—authenticity more than anything. It rewards many things, but one of the things it rewards most is authenticity. You'll see me on Twitter every once in a while make a genuine comment like, "Hey, thank you. I really respect you." People love to see that.
--:-- Guest (Bryan Kim): I saw your response to Gary Tan.
--:-- Guest (Roy Lee): Yeah, I mean it's true. I respect that guy. I hope that one day when the company gets to where I imagine it will be, he will come around. I don't think the lesson is to triple down on everything. I think the lesson is that if you're honest, the algorithm will reward you because there's literally zero other company out there that is being fully honest about everything. There's zero founder out there that behaves authentically. The only other person I might think is Elon Musk. I think that's a really good role model to have in business making.
--:-- Guest (Bryan Kim): As you build Cluely into the first destination for consumers and enterprise alike, how do the type of stunts and skits actually fit into the type of customer you want to serve?
--:-- Guest (Roy Lee): I think we're headed towards a future where this is the new normal. 40 years ago, we were way more professional than we are now. If you were an engineer, you came into work in a suit and tie. If you didn't, it was distasteful, disgusting. You should never show up in a hoodie and sweatpants. Now you're weird if you come in a suit and you're not in a hoodie and sweatpants. Everyone wants more authentic things. Right now, for some reason in society, we're still lingering onto this image that companies need to be brand friendly and boring and never say anything controversial. I don't understand how this became the societal norm.
--:-- Guest (Roy Lee): In reality, people want to see interesting things. That's the point of life—you see interesting things. There's this lack of professionalism. With the distribution of short-form content, everyone sees the craziest things. You get desensitized. That's why things like sperm racing are able to be so hyperviral. Nobody would have aired that on CNN 10 years ago. But you don't need CNN anymore. You have Instagram, TikTok. And Instagram and TikTok love sperm racing. As a result, they're able to raise at a massive valuation. They genuinely have a shot at being the next legit sport.
--:-- Guest (Roy Lee): It's not just sperm racing. It's every company in the world. There's Sam Altman talking about how hot the guys are that GPT generates in the timeline. There's Elon Musk doing ravings about his political takes. Every company is getting more controversial, less professional. This trend is not dying anytime soon. I'm just the person pushing the envelope perhaps a little further than the world is ready for at the moment.
--:-- Guest (Roy Lee): The question everyone should linger on is: can you imagine a world where we do win? What will the state of the world look like if Cluely wins? If we prove to everyone that the bar for professionalism is here, where we've determined it, the corporate culture of America as a whole is going to shift—perhaps the entire world. Everyone will realize, "Oh, we've had our panties in a bunch worrying about brand image and professionalism when in reality the world craves something different."
--:-- Guest (Roy Lee): I have very strong conviction that I'm right about this because I was right about X. I did not understand for the life of me why nobody is producing the viral videos that were so obviously designed to go viral for the algorithm. It's just because nobody has caught on and nobody's willing to press that button. Now that I've pressed the button once, imagine what the world looks like if Cluely makes it. That would be a much more interesting world if every company was being 100% radically transparent and doing exactly what the most interesting thing was. As Elon says, the most entertaining outcome is the most likely. It's true.
--:-- Host (Erik Torenberg): On that note, this has been a fantastic episode with Roy and Bryan. Thanks so much for coming on the podcast.
--:-- Guest (Roy Lee): Yeah, thank you so much for having me.
--:-- Guest (Bryan Kim): Thank you.
--:-- Host (Erik Torenberg): Thank you. Thanks for listening to the a16z podcast. If you enjoyed the episode, let us know by leaving a review at rate this podcast dot com slash a16z. We've got more great conversations coming your way. See you next time.
