I built a $70 Million app using this Easy Viral Strategy (copy me)

Macarron was just a test, even viral.

Today, we are speaking to KJ Chen, who is the founder and CEO of Macron, a personal AI agent that helps users build many apps with [music] just one

sentence. The company is already at a

sentence. The company is already at a valuation of $70 million after having to pivot from being an AI novel platform.

We had 3 million users. It was one of the biggest in the world. Don't you feel like that was a very large risk you were taking in terms of like potentially losing all your users?

It was a very big decision, a tough one.

He has now scaled to more than 300,000 users using viral marketing with short form videos and 200,000 apps created using his product.

The structure is post a problem in your real life and then a solution which is a mini app and then the code which you can download and enter and get that mini app. In this episode, KJ reveals his

app. In this episode, KJ reveals his exact viral marketing playbook, the briefs and the frameworks they used, and the step-by-step strategies he implemented to go from zero, a,000 to

20,000 and then to 100,000 and beyond doing the viral techniques, but also integrating your product a little bit better. Which one converts better?

better. Which one converts better?

It's usually the one that raise a Let's break down his startup strategies.

Welcome to the show, KJ. Who is KJ Chen and what do you do?

I'm the founder and CEO of Macron, which is a personal agent. It helps with your personal life by building mini apps for you. So, you can create a calorie

you. So, you can create a calorie tracker, you can create exercise tracker, anything you want. I graduated

from Duke and had many startup experience.

Amazing. Amazing. I know Macaron is currently at a $70 million valuation, but you pivoted in order to get here from what I know. Can you quickly tell the audience a little bit about your first version of your product?

Uh yeah. Um first I cannot comment on the valuation but uh before Macron we did another product called Midre. Uh it

was an AI novel writing platform. It had

three million users. It was one of the biggest AI novel writing platforms uh in the world. From the very beginning when

the world. From the very beginning when we started this venture it was in at the end of 2023. We began with a very strong thesis that we need to do research and product at the same time to provide some

amazing experience and at that time it was by training model we can write novels just like feels like human writing the novel so it's of great

quality and people come here to write um then we had a pivot in 2025 at that time we saw an interesting new opportunity

that is building apps for everyone I I know this sounds very disconnected but from AI novel to building apps for everyone is crazy. The connection was

that I was talking to a lot of uh mid-real users and they said that it was great using the product writing novels immersed in those stories but they have

to go back to real life again. When they

do that they feel empty and they they have to go back to study or go back to their work or have relationship with their family. I thought at that time why

their family. I thought at that time why not help people a little bit better in their personal life uh not just creating stories for them. So that's the connection that we made and our

technology also supports us uh in this transition to creating apps for for everyone and I think we were the maybe the first one for consumer to create

apps. We released our macron in in

apps. We released our macron in in August 2025 so it's pretty early.

Very nice. I must say, if you already had three million users, don't you feel like that was a very large risk you were taking in terms of like potentially losing all your users and like building out a fresh new product?

Uh yeah, definitely. Um it was a very big

yeah, definitely. Um it was a very big decision, a tough one. Um what I thought was uh we were really small uh so we

could only afford to do maybe one very specific thing. We were not able to

specific thing. We were not able to support two products at the same time.

At that time, Macarron was just a test because we train a model to write mini apps. It was just a testing out there.

apps. It was just a testing out there.

It went viral and we quickly gained uh tens of thousands of users through this test and we saw the momentum was uh much

stronger than midre. Uh because novel writing um is very hard to train for for each novel. you you only get fragmented

each novel. you you only get fragmented signal for for improvements uh in the reinforcement learning process. So we

think okay maybe mini app because people will click on a lot of buttons and when they create one they will give feedback that's a stronger feedback than uh reading or not reading a novel. So I

think that's a stronger moment and we shift the whole company into this new road. Yeah.

road. Yeah.

Okay. Okay. Okay. So, it seems like some of your early users kind of gave you feedback to dilute some of those concerns and risks, which is great cuz otherwise I was going to say that's like a massive risk you're taking in terms of

like the 3 million users being gone. Um,

but back to the current day with the time period that you guys have been around, I would love to know how many users you have right now.

Uh, we have more than 300,000 users right now. And then in terms of apps

right now. And then in terms of apps created on the product, how many is that? Currently

that? Currently people have created more than 200,000 apps on Macro.

Very nice. Very nice cuz that equates to almost like each user going out there and like uh creating and testing the product which is awesome to see. So

those users are very active. Then

the the statistics feels like a third of the people that h that is active is creating mini apps. On average, every one of them create two. So, adds up to

200,000.

Amazing. Amazing. Um, those are awesome numbers already. Let's kind of break

numbers already. Let's kind of break down how we ended up growing to the current numbers we're at about 300,000 or more um users that we have. I would

love to know maybe from your zero to,000 stage of users, what were the methods that you kind of had to use back then?

We were able to find some communities that were interested in creating creating small apps. Uh those were like uh AI maker community uh people that

were so interested in writing different prompts and prompt community which a prompt is basically a easier or simpler version of a app. And the best thing of

Macron is that you can create an app with just one sentence with just chatting. It was perfect for them to on

chatting. It was perfect for them to on board and create their own apps and it was super easy. Uh the community had about uh 500 600 people around different

friends just get into those community and share the idea. Uh they loved it. Of

course they gave us a lot of feedback because in in June and July our algorithm was pretty terrible. It can

create apps with a lot of fake buttons that can't click. It was a lot of trial and error that we we um tested at that phase. Yeah.

phase. Yeah.

Okay. There definitely was some hiccups then. Um it seems like though the

then. Um it seems like though the initial users were mostly just organic from whether it was like the network you already had as well as it seems like from the previous version of your

product as well. People started

trickling in from there. Okay. And then

the community testing you did.

Yeah. Yeah. previous product as well, but mainly it was through um prompt communities that I found.

Okay. Awesome. Awesome. Um I'm curious then what did the launch numbers look like?

It was cool. Um first we launched on product hunt. We were very lucky to be

product hunt. We were very lucky to be the the number one product of the week.

The product of the week and that um brought us some some traffic and I think it was because our product look very different from other ones. our

logo we're called macron and it really feels like macaron when you see the uh icon of the app. Um it was it was orange and pink and and very tasty and we also

released the videos which I think were uh quite popular among some uh tech communities because it was the first agent for people to build their uh small

apps for consumers to be there. Of

course at that time we already have lovable and uh bolt and other products but they were for developers and product managers mainly were for and consumers.

So we have around 10 20,000 on boarding users in the two weeks that follows the launch. So that was a very good boost.

launch. So that was a very good boost.

Yeah, I must say that that probably would be um some decent numbers actually just because I've heard of some terror stories in regards to product hunt. like

people think it would do super well and then some people like just get really really flat results and like no movement at all. But that is great to hear. Um

at all. But that is great to hear. Um

I'm kind of curious then from there your first 20,000 users now that you have them what is the method that you're using and the strategy that you're using

to get to the 100,000 users and 500,000 users and beyond? We started a lot of um video campaigns and that video campaigns

feels like uh of course Cluey is is the best in doing this or uh the OG in doing this but uh we copied

what KJ is referring to here is the marketing strategy that was spearheaded by Roy Lee the founder of Clearly the AI powered desktop application that acts as an invisible realtime assistant during

meetings calls calls and interviews.

What was unique about Royy's approach was that he aimed for mass reach over targeted reach. Clearly, his strategy

targeted reach. Clearly, his strategy can be broken down into mass content production, producing hundreds of videos per day at their peak, their viral

formula. If you posted 100 videos on one

formula. If you posted 100 videos on one account, you may get virality. If you

then repost this on other accounts, you'll likely achieve a multiplication effect. Number three, controversy as

effect. Number three, controversy as distribution. They aimed to polarize

distribution. They aimed to polarize their audience to stir reaction and engagement. Hiring a creator army. They

engagement. Hiring a creator army. They

used 50 plus creators to produce multiple different videos a day. And

last but not least, a distribution first approach. They just hired engineers and

approach. They just hired engineers and creators, no inbetweens. And this

resulted in more than 1 billion views in 3 months and a 7 million ARR with a $120 million valuation. We reward people for

million valuation. We reward people for recording their mini apps and recording their activity in their mini apps and share share them on Tik Tok and other

platforms. And many people did this and they're not sharing the ideas that we originally thought they would. For

example, creating a calorie tracker. Uh

they they usually create something that's a little bit more absurd and get traffic online. For example, they create

traffic online. For example, they create a exercise um planner, but it makes crazy exercises and the the influencer

just follows the exercise plan and in the end he laid lay down on the on the on the floor and feeling too tired and he he just made a video of that whole

thing and it went viral and gained more than a million views on Tik Tok. things

like these. Um, so they they are not predicted but but we made a lot of videos in that way and the traffic start to grow from there. Again,

that is super interesting. So it's kind of highlighting almost the product in a very authentic way where it doesn't necessarily say the product is great, try this product. It's actually kind of

talking about a very funny circumstance that came through trying to create an app. I'm kind of curious, is this

app. I'm kind of curious, is this something that actually converted to users or not really? Before we dive further into the episode, I wanted to share something that I use for growth hacking. If you want to find your

hacking. If you want to find your competitor strategies and outsmart them, this is exactly the platform for that.

Let's imagine you have a generative AI startup that helps people write copy for different social posts. We want to go on similar web and list all your different competitors. This web analysis feature

competitors. This web analysis feature will find for you which site is actually getting the most traffic and using that you can also find which channel strategy is actually leading to those results.

Additionally, given the day and age, you also need to be making sure that you're popping up on chat so that you can win the next gold rush that is AI SEO. You

can track any website with the AI brand visibility tool. Then with prompt

visibility tool. Then with prompt analysis, you'll see what people are asking AI about your product, category, or industry, and whether your brand is showing up or not. They currently have a

7-day trial. And if you like what you

7-day trial. And if you like what you find, you can use the code Juliana 20 for 20% off your subscription each month. Try it now in the link below. So,

month. Try it now in the link below. So,

what the what the video does is that uh people can leave a code for for the app that they created. For example, if you create an app, you can share your code

in the video and people with that code can instantly get your app in Macron. Uh

this is our sharing feature. Um which I think is very useful. Uh people people love it. When they adopt the app in this

love it. When they adopt the app in this way, they you will usually uh check the code first and check to see what that influencer did in the Tik Tok video and

whether that's fun or not and and other things. So uh when you have that code uh

things. So uh when you have that code uh you naturally start chatting with macron and then from there macron will discover more of your needs in your personal

life. Uh whether it's your frustration

life. Uh whether it's your frustration of dealing with different difficulties or your curiosity of exploring different things like maybe um studying tennis or

new sports, new instruments, other things and can create um mini apps for you to learn those things. So it will naturally grow once you have a contact point.

Very nice. So it seems like you have your viral marketing framework and then supporting that is almost like a productled growth um framework that is

also bringing in more um users into the platform. I had a look at some of your

platform. I had a look at some of your videos and saw a lot of different like viral videos, right? I've seen random ASMR videos. I've also seen people Yeah.

ASMR videos. I've also seen people Yeah.

talking about the different apps they're creating, whether it's like workout calorie counters and stuff like that.

I've also seen lots of like tech influencers kind of talk about the product as well. I'm kind of curious how you ended up picking those types of creators out or was there a particular plan around it or it just seemed like

those were the ones that were working. I

I think the ones that you can see are the ones that that's uh selected by the market. So, uh I I don't usually pick

market. So, uh I I don't usually pick too much. Um so anyone who wants to

too much. Um so anyone who wants to create a video for macron I'll just say go for it and if your video creates a lot of view I will pay for it. Uh so

it's very simple and um of course I I may want uh it to be more personal life oriented instead of creating slides or

power powerpoints of uh reports. Uh

that's the productivity direction which is not background. uh but I don't usually select um and it's it's great because um if people are creating videos

that thank you for everyone who created video but anyway if the video doesn't gain much traffic I don't need to pay that much anyway so so it's uh yeah pay

pay by myself very nice so it seems like then they're almost like affiliates and these are the people who are kind of pushing the initial strategy forward in terms of

like viral marketing for you. I am kind of curious then because I've seen such an array of like different types of creators out there. Which ones are the ones that are converting best? Cuz like

I just went through a couple, right?

There's like random ASMR ones doing storytelling, tech influencers, like young adults doing workouts and stuff like that. Is

there a particular category that's working the best for you? Usually if I found a video of high quality and I think this is good, this is a representative of our product, uh the

video usually doesn't do well, uh maybe my page is too old. Maybe maybe I'm thinking too much from the product side and not from the uh people browsing

videos that that angle. So at this point I I just gave up in determining which video will go viral. Okay, that's super interesting because I was going to say there's different types of virality,

right? And what I noticed is there's

right? And what I noticed is there's virality for reach, which you're just trying to reach the largest number of people. It doesn't necessarily talk

people. It doesn't necessarily talk about your product too much or doesn't necessarily try to sell the product too much. And then there's like the mid-tier

much. And then there's like the mid-tier virality, which is doing the viral techniques, but also integrating your product a little bit better. Like I've

seen some of the tech people talking about your product in a little bit of a more how-to style manner, right? And

then there was this funny video of like someone talking about um what GPA you got and then they were kind of like singing along to it. And I'm [laughter] kind of curious which one converts

better. So, in terms of the the format

better. So, in terms of the the format or the flow of the video, it's usually the one that raise a real question or

situation in your life and offers a mini app for that. And I have a code.

Let's go into that viral format in a little bit more detail and how exactly to optimize it for conversions. These

formats can be broken down into three core steps. Number one, introducing a

core steps. Number one, introducing a problem. We see here that Macaron's most

problem. We see here that Macaron's most successful videos introduce a relatable situation for viewers that draws the attention in. Number two, it's only then

attention in. Number two, it's only then that we introduce the solution.

Formatting the creator campaign videos in this way allows for a more seamless integration of the product where we show the benefits without having to sell it.

Then three, we end with a CTA. In this

case, a custom code.

The the GPA video. I I know which which one you were talking about, but it was pretty pretty fun, but people don't have a strong incentive to to to download the

app. So, the conversion is is not that

app. So, the conversion is is not that good. But, uh it was okay, but not that

good. But, uh it was okay, but not that good. But, um for the ones, for example,

good. But, um for the ones, for example, exercise and you need to download and enter this code to get this exercise and you can feel that um that that will have

pretty good conversion. The structure is post a problem in your real life and the viewer does have that problem and then a solution which is a mini app and then the code which you can download and

enter and get that mini app the solution I provide you. So that's the general flow.

Amazing. Thank you for saying that because I always try to tell people as well that like viral marketing and just aiming for mass reach and like it's not going to convert to users and you need

to like formulate your virality to actually get users on board and make the piece of content make sense whilst aiming for virality for it to lead to

users. So appreciate the confirmation.

users. So appreciate the confirmation.

Ju just to add to that um I I always feel that we need to give the creative freedom to um everyone that that post content and I do and but when I say uh I

I like the absurd ideas or free ideas um I I'm saying that it's still within this framework that it works best for our product. It's more about I don't choose

product. It's more about I don't choose which mini app you want to create.

people will create uh apps that make memes and maybe cut photos. Uh these

things are things that I would not create but uh if you choose that that's great that you still need to follow the problem format. Yeah.

problem format. Yeah.

Amazing. Um more on that I would kind of love to break down with you then. Is

there a particular brief that you give to these creators in terms of like follow this kind of framework but you do it in your way and how you would integrate it into your life or um do you just

just point out like a few points and then creative freedom is up to them.

Yeah, I will give them uh a brief of course about what the product because it's still pretty new. Um in terms of the concept, most people haven't used

anything that can create apps for them.

So the concept needs to be laid down.

But the brief is pretty uh concise. I

would say 1,00 200 2,000 words and explain everything with a lot of pictures and links. We have tears mainly because we are aiming different

directions from time to time. Sometimes

we we need to show our technology more for the community and mainly for hiring people. Uh and that's a very different

people. Uh and that's a very different brief from maybe showing our product side.

I know that's a lot of information probably mainly for like Tik Tok and Instagram like the different short form pieces of content. I would love to dive into X, however, because I know you've

gone viral on there multiple times as well. Do you have like a different

well. Do you have like a different process that you do or is it like similar in terms of like creating virality on there?

So, I I think I'm not that viral on X.

Uh we [laughter] still need a lot of work. Um I think our most recent video have 3 million views, but uh it's still quite small compared

to other many other people. Anyways, so

uh what we basically do on X is that we we still found those communities and influencers that may like us or that we can cooperate. For example, our I think

can cooperate. For example, our I think our second launch on on X was more successful which was at the beginning of December. Uh it was a tech released

December. Uh it was a tech released about our new technology in mind lab which is the lab that supports Macron.

Uh and in that lab we released a few works which is the reinforcement learning infrastructure that allows us to train the really large models the

trillion size models and we release our memory algorithm um which is really cool because it was uh uh state-of-the-art performance and really impacts the

product because uh every every chat is uh cooked with memory. We released these work and we found uh people who have

already show an interest in these work on X and we direct message them uh saying that we have this new release coming up and this is our work. Uh it

feels like um talking to other researchers and and showing our research results. If you want to uh do a rebuttal

results. If you want to uh do a rebuttal or uh show show your insights on this work, please do. Um so we DM'd a lot of

people uh in this direction and naturally they show uh great interest because our work is it's it's still pretty good work and uh the work has

also contributed to Nvidia and a lot of open source projects. So you're very interested and they accept that they share the work.

Definitely. I mean, what's good to see about what you've done there, I think, is that you've kind of leaned into platform and creator fit and you're kind of looking into, oh, what is most

suitable for the second product launch because obviously it's a little bit more um I guess more suitable for people who are a little bit techsavvy. So, you kind of shifted towards X, which is kind of good to see so that you do have that

fit. Um, but I'm curious, what do you

fit. Um, but I'm curious, what do you think has been most successful for you?

cuz I cuz your product is kind of at not only is it like technologically savvy, but it's also something that a lot of consumers can utilize. I'm kind of curious which platform has been most

successful for you in terms of actually driving the users that you want to see.

Tik Tok and and Instagram. Uh these are the the big ones and Tik Tok might be a little bit uh cuz uh it's easier to

create a video loop on Tik Tok uh for us.

Um I'm kind of curious then what are your next maybe you shouldn't give these away but what are your next plans in terms of like your marketing strategies that you're looking to implement further

to kind of get to your 1 million 2 million user mark?

Yeah. Yeah. Yeah. Um the there are a lot lot of things that we need to do we want to do and a big one of them is our website

um building SEO from there and this is because um of course we had an app uh that's for sure but many small tools we

found people created can actually be shared online and um actually for one-time use or for search engine use.

This is important because uh it's basically free traffic because we already have um 200,000 apps and many of

them can be converted into a single website app and those app can if each one of them gain 10 traffic uh each day

we already have 2 million traffic which is great. So that's one of our strategy

is great. So that's one of our strategy moving forward. So marketing is always

moving forward. So marketing is always connected with act what what product feature you are launching and we are launching something very new uh in terms of experience may not have been seen on

the market before like when we launched macron it was a good good launch uh we want to recreate that launch and maybe people will come from there uh it was a

research product code design experience that that we have high hope for okay I'm excited I'm excited to see um what comes next. I think I'm also

curious, are you guys planning to like continue on with I guess scaling up your viral strategies that you're implementing currently?

Um definitely. Yes. Yes. Yes. Yes. Uh I

I think it will grow with our number of users as well. Um, so we aim to have maybe maybe 200,000 or 300,000 daily

active users uh next year, which is a 10x growth from uh what we currently have now. Um, so we definitely need to

have now. Um, so we definitely need to scale this. Right now our marketing team

scale this. Right now our marketing team is also very small. Uh, of course I I want it to be more organic of course.

Um, but I I think it will scale up as well. Um, yeah, definitely. Let's go.

well. Um, yeah, definitely. Let's go.

Before we let you go, I would love to know if you have any final words for anyone else building like BTOC products out there and some tips you might give to them.

Okay. Um, I I'm not in a position that I can give advice. Uh I think I'm still fighting this war and um the only thing

that I found um worth daring is that uh we are very different in terms of we trend our own model and we always want to give some experience to user that's

special and never seen on the market before when you go into this app that's that's very different from what you experience and I think you can achieve that with prompt engineering of course

or context engineering um or training a model which is becoming easier and easier today. So um yeah, if you are

easier today. So um yeah, if you are creating something that's for consumer, I think people do want to experience something that they've never seen before.

I appreciate the tips and the strategies you've shared. Um and thank you for

you've shared. Um and thank you for coming on the show.

Thank you. Thank you for inviting me