18 Year Old Got Paid $2.9m Last Month for New AI Mobile App
20:19 Share

Are you comfortable sharing how much revenue you've already collected this year? A little over 10 right now. And I mean, how many people are running? How many people are downloading it right now per day? What's your sample size every day? Every day we get

20 to 30,000 new downloads. It's wild. How are you thinking about a user going into their second year and the activation rate, the renewal rate? We just came around our year anniversary since launch. However, we expect that it's going to be between 25 to 50% renew. What is the team size today full-time? How many folks? We are 17 people actually. Hey folks, I'm here with the founder of Cal AI, Zach Yaragari. Zach, you ready to take us to the top?

Hey, thanks for having me on. Yeah, I'm thrilled you're here, man. You know, you're really good at press. You've built a great story. You're a great engineer, but you're even more powerful sort of marketer. We're going to jump in today to the story of Cal AI. Just tell us real quick if for folks haven't if they haven't seen the app yet. What does it do? What's the main use case?

Cal AI is an app where you can track your calories just by taking a picture of your food. So it's the easiest way if you're trying to lose or gain weight. You were on your other interviews talking about how you're a skinny kid and it's really hard to like measure every fish that you put in your mouth or every meal. And this just makes it easy. How much of the, of the code, I mean, is this, you know, do you consider yourself a really, really good engineer? Was this a hard breakthrough that is only possible in the world of sort of AI or could this have been built five years ago?

I think it definitely is only possible now because of the breakthroughs with AI. However, that's mainly because of APIs like ChatGBT. We did not

portraying our own model to do this. Yep. Yep. Okay. That makes a ton of sense. So a lot of things I want to touch on with you today is just growth. So I don't want to lose the audience. I don't want to bury the lead, right? This is your screenshot you posted back in September 4th, 2024. When you guys, this is your first million of sales, right? Yeah. That was like five months into the company. That's wild. Okay. So first line of code was written what in like May, 2024?

We launched May, 2024. The first line of code was probably written in March. Okay. And solo founder here or multiple co-founders. How'd you guys get the company going? Started as three co-founders. Two of us were working on it. The third was more of an advisor. And then we brought someone else on. Everyone became more hands-on as it started taking off. And is this the crew? Is this the right picture or no?

So that's three of the co-founders. I'm on the left. Henry is to the right of me. He's our CTO. And Jake is all the way to the right. He's our COO slash CMO. But Blake is the fourth co-founder. He's not in that picture. Okay, got it. That makes a ton of sense. So if we fast forward a little bit, again, growth has been pretty impressive. I cheated here and used perplexity, but why not? Might as well use the tools, right? You get going in May, two folks. You scale to six people. I'm really focused on this concept of like tiny teams and huge revenue. Are you comfortable sharing where you're at?

today in terms of revenue and team size? Of course. Today, in the next 12 months, we're looking to generate over 35 million. That's just based on our annual run rate. So just to be clear, you think in 2020, we're recording this in May 2025, you think by December 2025, total collected revenue in 2025 will be about 30 billion? So that's assuming no growth, but counting for growth, it should be higher. Okay. Are you comfortable sharing how much revenue you've already collected this year?

So this year alone, we have collected something like six,

- Something like nine million. - Nine, okay great. That's pretty good over the first five, six months of the year. So good growth. What I'd love to jump into is a little bit about how you actually architected the application. So I went ahead and went through as a user, recorded it here. I'd love for you to just take us through this. So someone's searching calorie tracking in the app store. By the way, I guess we should probably start there. Do you do any sort of, like the equivalent of SEO for Google, but sort of meta tags in your app here to show up for these key keywords?

So we do, we have actually done some things like our keywords. You could definitely see through our name and the subtitle. We definitely try to fit in as many keywords as we can. What tools are you using to figure out what keywords are the most searched on the app store so you can then rank for them? So tools, we actually don't use any tools. I know app figures is a very popular one, but we mainly use, it was mainly off of the dome actually. What's the dome?

It was mainly off the top of our minds. Come on, man. I'm 35. You're like a young guy. Like all these, all these phrases I'm still catching up. So, okay. So you just came up, you just said, Hey, listen, I'm, I'm the power user. What would I search? Let me go rank for those terms. Yeah. That's how we did it initially. And,

Going up a little bit, we started to talk to some agencies, so had more of a clue, but yeah, definitely most of that is just what our intuition told us. - Yeah, so let's get into the app here. So I'm just scrolling through here. You got Viral Dental on LLC, sorry, the holding company. You told me that's not the strategy, multiple apps like CalAI is just crushing it, that accurate.

Yeah. So the goal initially, basically we saw, we thought that our one, the one thing we had that other people didn't was our influencer marketing strategies. We had hundreds or we have hundreds of influencers that are promoting our app posting about it every single day. And so we, the plan was to apply that same thing to other apps.

But Cal AI just kept growing faster and faster. And so by the time we had finished development on those other ideas we had, we realized it didn't make sense to spend any time there. Just given Cal AI's growth rate, it deserved all of our attention. Yeah, that makes tons of sense. And I meant to ask you earlier too, so 9 million collected so far this year, original team, you just walked us through it. What is the team size today full-time? How many folks?

So I was backtracking on the revenue calculations, probably closer to like a little over 10 right now. And in terms of the team size, we are 17 people. Actually, we're bringing some people on now. So we're moving from 17, which we have stayed at for the last few months, to we'll probably be at...

22 people by the end of the next month or two. And how are you finding some of these folks? I mean, I saw on Twitter somewhere someone posted like just sent you new screenshots of the app and said, hey, I could make the conversion better. And you said, OK, like join the team. I mean, how are you recruiting folks? Is that easy? I'd say most of our talent came straight from X. So as you mentioned, two days ago, someone redesigned a Kali screen and I

I like the design, so now we're interviewing him to join the team full time. I made a post the other day, yesterday actually, looking for someone to come on as a Facebook ad buyer for us. Yeah, this is the easy way to do it. Not the easy way, but it's a quick way to do it, right? These are the hustlers posting real-time stuff here, instantly trying to capture your attention, which is great.

So let's keep going through the app experience, right? So how much testing did you guys do in sort of the first five screen sessions and what should we show? What GIFs should we show? What's too fast? What's too much, et cetera? - Are you asking about the onboarding flow specifically? - Yeah, let me just actually here, I'll just make this easier. I'll just actually share the parts I'm specifically curious about. So yeah, this onboarding flow here, right? How do you decide what GIFs to show, what headlines to show, what to put in the onboarding flow here?

Yeah, so that's a good question. Initially, we had no animations. We had no GIFs. It was very bland. We only asked the questions we needed in order to give the user their calorie daily intake estimate that they would need to gain weight or lose weight. And

That was, it was purely utility. But then over time, we started A-B testing different things. We started adding in different questions, which made the user spend more time. But these questions didn't actually have an impact on the end result at all.

But we saw an increase in conversion rate and the hypothesis there, which is pretty confirmed. A lot of other people see this is just that the user invests more time. So something like following a specific diet, it doesn't impact their app experience at all, but that actually boosted the conversion rate, adding these questions. And then we add the idea to, okay, let's add a screen that says, thank you for trusting us. Let's add a screen that says you're going to lose weight faster with Cal AI.

And that was cool. That increased the conversion rate. So how do we take this one step further? We animated those screens, tested it again, also boosted the conversion rate. So it's just been a game of creating hypotheses and then testing them out.

Yeah, I mean, I joke with folks. I say, listen, whichever teams are moving the fastest and hustling the hardest, the one that gets the 10,000 tests the fastest will be the winner because you're learning the quickest. And so are you following any testing framework on this stuff? Is there a software powering this question loop that we just went through? Is it all sort of custom coded?

It's custom coded. We use Mixpanel for our tests right now, and I'd say we don't do anything that crazy. We also are definitely not testing as much as I would like to. The 10,000 tests, I would love to run 10,000 tests. We've only started running tests since maybe January, three months ago. We are...

behind on that front and have mainly operated through intuition. But now everything, every new feature we release, we are running to us. - Yeah, I mean how many people are running, how many people are downloading it right now per day? What's your sample size every day? - Every day we get 20 to 30,000 new downloads. - That's wild. Okay, so we're setting it up here. I've already invested in the app 'cause I just spent a minute telling you my weight, my birth date, like all that kind of stuff. How many people convert from this experience, like the $30,000 per day, right into a paid plan?

So we have a three day free trial. Not everyone goes right into a free plan. I'd say into the trial, it's about 25%, 20 to 25 will convert either to paying right away or into a trial. That makes tons of sense. You guys do like this trial process here is so fricking smart. Like when I was clicking through it, I'm going, this guy, it's a fricking genius because you have this continue for free button here. Right. And it goes two 49 per month. And I'm like, of course I'm going to do a three day free trial. It's 29 per year.

But then what you're thinking is you don't just like, you don't just lay back on your laurels at that point. You then actually upsell me here and you say, wait a second, I've got a code for you. You got to use it today. If you use it today, it will take you like directly into a paid plan. Whose idea was that? Is that software? Do you have a, is there a tool that's powering that or is that something that you guys built from scratch? We use a tool called super wall right here, our paywalls. And that's actually worked out really great for us. Can you quantify that? This is super wall right here, right?

So this is all built through Superwall, yes. It's a web paywall viewer. It makes it super easy. And so you, true or false, once you installed these, you saw higher conversions, more people paid on the first session.

So you have to build the paywalls. Well, they have templates and then run the test, but 100% our conversion rate is much higher for using Superwall. It has allowed us to run so many tests. That's the one thing we don't lack on. We definitely lack on testing features, but testing out paywalls, we're running so many tests on that. Here's what's crazy. It's very different than B2B SaaS. You don't ask for the signup until all the way at the end.

Like all the way at the end. That was obviously intentional. You got it, man. Yeah. What other things did you eliminate from like traditional onboarding flows so that you could have less friction?

That's a good question. Well, early on we had like nothing, but then we saw that adding more questions would increase the conversion rate. Definitely moving the sign in screens to the end because there is a lot of drop off there. That was the biggest by far. Okay. That makes tons of sense. Look, we've covered a bunch of stuff so far. We talked about onboarding flows. We talked about pricing. What else is top of mind for you? Anything I should have asked that I haven't asked already?

You're asking some pretty good questions. Pretty spot on, I would say. All right. All right. Cool. You're doing the whole university thing. You want to spend that time with other folks your age. Like, I totally get that. It's also potentially a great place. You're going to end up hiring professors, I think, right? You're going to go in there and recruit the head of CS.

Yeah, I mean, we'll see. You're not just someone that's sort of, I don't want to say really good at social media here. You also know how to take an acorn piece of content that might go viral on Twitter and translate that into traditional media, cable appearances, things like that. That doesn't just happen by accident, right? How have you done that? Well, I had this crazy moment on Twitter going viral about my college admissions. I knew it would go viral.

I'm kind of wired to know when posts will go viral just because I grew up on social media. But I thought that it would get like two or three million views. It got 28 million views. And then if you scroll down, the personal statement got 40 million views by itself. And so that garnered so much attention. And we actually had a few stations reaching out.

early on it was like one or two but then as I made appearances more started reaching out so it's definitely a snowball effect here where once you're seen on one the other stations probably get some FOMO they see okay this guy actually can present well on camera we should bring him on our show too. Now it makes sense that what I appreciate about you is there's a lot of people that are sort of hustling and stuff and they just say provocative stuff that's just like it hurts somebody else when they say it you're going viral in ways where you're just a genius storyteller you're not

knocking someone else down while you make the post, right? And so this is just proof that you can get attention. You can build a great media brand around your company doing it the right way. And so this is great. I'm excited for you. Excited to see what you do here in college. Anything else that I want to make sure? So we touched on the early team here. We touched on downloads. We touched on conversion rates. The team's remote, I assume, right? Everyone is remote. Maybe we should touch on marketing a little bit. Let's do it. What's top of mind for you on that right now?

So our biggest way that we've marketed the app, our biggest channel is through influencers on both TikTok and Instagram. We have about 250 influencers right now on retainer that are just posting, integrating us into their content monthly. So this account right now,

I wouldn't actually say this. I'd say go to our Instagram and then go to our real section and you'll see our collab posts with other influencers. - You mentioned something interesting though. So you find, let's say I'm an influencer with a big audience. You would reach out to me and say, "Nathan, I'm gonna pay you a thousand bucks a month, "fix, just make two posts per month." Or walk me through, like let's negotiate live how you do with other influencers.

So we would definitely start by making an offer like that. We would first look at your account is first and foremost, we would look at your metrics, your views, but then not just views because not all views are valued the same. We'd look at your engagement and based on all of these small factors, we would come up with what we think is a fair price, the maximum and minimum that we would want to pay you for your, for your posts.

- Interesting, so can you share maybe what's the, you don't have to name the influencer if you don't want, but what's the largest influencer? What do you pay them per month and what's the expectation that they deliver, number of posts-wise for that? - Per month, the largest, it's in the tens of thousands per month and the expectation is four posts. - Interesting, and do you give them any guidance on that? You say, listen, just be creative, figure out a way to mention CalAI.

They are the professionals. They know what goes viral for their audience. So our best, our job is to make sure they integrate the product well. But in terms of the video concept, we'll leave that up to them. Interesting. And how many influencers are you paying at least a dollar in fixed fees per month? We are paying 250 influencers right now. Wow. Okay. So 250. And so, I mean, is that just one-on-one recruiting or do you use an agency to go find those folks?

That's why we've had to build out kind of a I mean it you say this is small teams podcast I feel like we're a big team at this point not that big I suppose but 17 people about half of those are just on the marketing side right now working with Wow and and these influencers why not go why not only pay them for when they drive you a customer why go that why ends like a commission structure I pay them a flat fee instead that would be the dream but

But the reality is that there's too much competition out here with the brands and they want to be guaranteed a payment. And if we're not going to do it, another brand will. Yeah. Yeah. That's super smart. It's limited resources. It's not, it's scarce. Totally. Yeah. Interesting. Can you, are you comfortable? Could you quantify like last month, how much total you paid out to influencers? Six figures, six figures. Okay. Like four or 500 grand, something like that. Mid six figures. Yeah. Okay. Is that something you're going to like double, triple down on assuming you can find enough influencers to pay?

So that's what we've been mainly focused on doubling down on the last 12 months. And that's, yeah, we've continued to grow every month. We're signing new influencers. Okay, very cool. Hey, Zach, I want to be respectful of your time. You want to talk for maybe four or five more minutes on other marketing channels or you want to wrap up?

I'd be happy to stay on for a few more minutes. All right. So we just talked about sort of TikTok. Sorry, I'm trying to get an Instagram here on my desktop and I don't remember my password, so I can't get to it. Is there any other way I could showcase or can you screen share something? I'm actually banned on Instagram at the moment. I think I'm being targeted by some other account. Wait, tell me more. I mean, why? People are jealous you're growing so quick or what?

Maybe. I think there's probably some aspect of jealousy. There's also a business aspect to this. I saw someone made a fraud account of me, and I'm imagining that they're DMing my followers, probably offering some course or something that's generating them revenue, and they're making enough to justify paying to get my account banned.

- Wild, dude. Art of war, social media edition, right? - Crazy. - This is wild. What kinds of things, are there any sort of ninja things you're doing that I haven't asked about besides the viral stuff, besides the influencers, you've got the referral code system, you've mentioned rapid product iteration in other interviews, is there anything you wanna touch on there that's sort of not common? - Rapid product, the main thing that's, I think product is,

It's kind of straightforward when you just talk about rapid product iteration. I mean, it's just about moving fast. And I think that's what most brands, most companies don't do well is moving fast. I think most companies move too slow. And that's actually our number one core value at Cali is speed above everything. It's push things out fast, develop it fast, pushed out fast, see the test results quickly so that we could get going, get more experiments running, move along very quickly. Something interesting we are focusing on, though,

is hiring. That is the number one priority right now is hiring, but not just hiring anyone, specialized talent. So my goal right now for the company is let's bring someone on whose only job is just paywall experiments all day. He's just thinking about how can I increase the conversion rate? Another guy, his only job is making the food scanner more accurate. Another guy, his only job is the homepage.

That kind of thing is what we're focused on right now. Really interesting. Super, super, super project specific. Are you using any of these sort of like gum loops or these MCP agentic interfaces to like automatically run any of these viral loops or is it all humans right now?

Right now, all humans. I mean, definitely some of the developers are using Cursor. Okay. Yeah. And I guess we should touch on retention here, right? So you do a really good job of driving people into annual plans. You really don't even act. It's crazy. You've done so much revenue. You don't even really have a year under the belt yet. But how are you thinking about a user going into their second year and the activation rate, the renewal rate?

That's something that we're starting to focus on now a lot. We don't have the data yet because we just came around our year anniversary since launch. However, we expect that it's going to be between 25% to 50% renew, which is very broad, but that's the window we expect it to be. Very cool, man. Well, hey, listen, as we wrap up, is there anyone else you just really respect that we should get some love to that are the tiny teams, giant revenue, young and hustling? Yeah.

I'll think about that and I'll send you some people. All right, guys, there we have it. Zach at Cal AI doing over 30,000 downloads per day, over 10 million of revenue collected so far in 2025 on track for 30 million. He says, you know what? I'm making a lot of money. I'm still going to go to school. I'm still going to have fun. He's focused on hiring viral loops and a lot of pricing paywall updates. Zach, thanks for taking us to the top, man. Thank you, Nathan, for having me.