YT YouTube2Text

YouTube2Text → Transcripts → ChatGPT Made This Broke Kid $1 Million a Month in 18 Months

ChatGPT Made This Broke Kid $1 Million a Month in 18 Months — Transcript

by Vaibhav Sisinty · 3,032 words · 464 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00This guy pulls in almost 9 crore rupees
0:02a month. That's around a million
0:04dollars. And a year and a half ago, he
0:06was dead broke, sleeping [music] in his
0:08parents' attic with his older brother
0:10sending him money just to buy groceries.
0:12Then in about 18 months, the whole thing
0:15flipped. [music] He built one app after
0:17another that millions of people now
0:19download and became one of the biggest
0:21young founders on the internet. And the
0:23wild part, he's not even an engineer,
0:25never built an app before in his life.
0:27He just figured out how to sit in front
0:29of ChatGPT and ask it the right things.
0:32So, here's what you're walking away
0:33with. We're going to walk through his
0:34entire journey and pull out every single
0:37thing you can use. It starts with his
0:39exact playbook. And it's basically three
0:41moves that he ran every [music] single
0:42time. And we've broken all of them down
0:44so you can run them on your own idea.
0:45Then the exact tools he used to build
0:47all of it so you can start even if
0:49you've never written a line of code.
0:51[music] And finally, five app ideas that
0:53you can start building today. By the end
0:55of this video, you won't just know how
0:57he did it, you'll know how to do it
0:59yourself. Oh, and one thing before we
1:01do, I'm putting his complete playbook,
1:03the full step-by-step version, inside my
1:05WhatsApp community. So, if you want the
1:07whole thing laid out to follow along
1:09yourself, the link's right there in the
1:11description. Go join it. Let's go back
1:13to where it all started. App one, Rizz
1:15GPT. The story starts in 2023 with
1:18Blake's college roommate. He was a nice
1:20guy, but was painfully single and kept
1:22getting stuck on dating apps. Whenever a
1:24match would come in, he'd want to chat,
1:26but didn't know what to say to make the
1:28girl like him. It's a tiny problem, but
1:30also one that millions of people face.
1:32Blake noticed [music]
1:33that his roommate did not need a dating
1:35coach. He needed someone to write the
1:37first message for him. That's when he
1:39had the idea. What if AI just wrote the
1:41reply for him? So, that's exactly what
1:43he built. And it's almost insultingly
1:45simple. You screenshot the conversation,
1:47the app reads the text off the image and
1:49feeds [music] it to ChatGPT. And it then
1:51sends back a few flirty options. He even
1:54built a little spice meter to dial up
1:56how bold it got and then he launched it.
1:58By every normal startup standard, Ris
2:01GPT was a disaster. Notifications barely
2:03worked. There were no reviews. He had
2:05accidentally exposed his secret API key.
2:08Most founders would have quietly taken
2:10the app down and rebuilt it. Blake did
2:12the opposite. He shipped it anyway
2:14because he understood something most
2:16engineers don't. On the internet,
2:17distribution matters more than
2:19perfection. So, instead of spending
2:21thousands on ads, he went hunting in a
2:23strange corner of TikTok. There were
2:25anonymous accounts posting pickup lines
2:27and dating advice pulling in millions of
2:29views despite [music] having barely any
2:31followers. They were the perfect
2:33distribution machine. He paid two of
2:35those creators $50 each, $100 in total.
2:39They slipped Ris GPT into their
2:40slideshows so naturally that it didn't
2:43even feel like an ad. It felt like the
2:45answer to a problem their audiences were
2:47already talking about and overnight the
2:50internet did its thing. Within roughly a
2:52week, Ris GPT crossed 200,000 downloads.
2:55The app that began because one guy
2:57didn't know what to text his Tinder
2:58match was suddenly making around $80,000
3:01a month. Years later, [music] it still
3:03generates close to $200,000 every month.
3:06Eventually, he renamed it Plug [music]
3:09AI after another company threatened
3:11legal action. But by then, the lesson
3:13had already been learned. You don't need
3:15to invent a new technology. [music] You
3:17just need to stand where a cultural
3:18trend and a technological shift collide.
3:21Ris GPT was never the destination. It
3:23was the experiment [music] to give Blake
3:25proof that he could spot an internet
3:27wave before everyone else and build
3:29exactly what that wave needed. And then
3:32he saw an even bigger one. If you've
3:33spent even 5 minutes on TikTok, you've
3:35probably seen it. Young men comparing
3:38jawlines, rating each others faces,
3:40posting glow ups, and asking complete
3:42strangers, "Be honest, how attractive am
3:44I?" An entire [music] generation had
3:46become obsessed with one thing, looks.
3:48The internet even gave it a name,
3:50looksmaxing. Millions of young people
3:52were trying to optimize their appearance
3:54through grooming, fitness, and [music]
3:55styling. And at that exact moment,
3:58another wave arrived. OpenAI had just
4:00released GPT-4 Vision. For the first
4:02time, an AI could look at a photograph
4:05and actually understand what it was
4:06seeing. Before that, building a
4:08face-rating app would have required
4:10[music] custom computer vision models, a
4:12large engineering team, months of work,
4:15and tens of thousands of dollars.
4:16Suddenly, one API could do most of the
4:19heavy lifting. And Blake recognized the
4:20same pattern [music] he had seen with
4:22Riz GPT. One, a massive cultural
4:25obsession. Two, a brand new technology.
4:27[music] And almost nobody was standing
4:29where those two waves met. Because he
4:31saw that software developers don't
4:33understand virality. The internet
4:35[music] money kids don't understand
4:36software. There's a tiny overlap, and
4:38that's where the opportunity is. So,
4:40[music] he built a startup called Umax.
4:42The idea was ridiculously simple. You
4:44upload a selfie, [music] the AI scores
4:46your face across more than 20 traits,
4:49and gives you marks. Then it tells you
4:51what can actually [music] be improved,
4:53like your skin, hairstyle, body fat, and
4:55grooming. Not genetics, [music]
4:57but things you can genuinely change,
4:59like your grooming, fitness, and
5:01styling. Just like a styling coach.
5:03>> [music]
5:03>> And it was built in very simple ways.
5:05Blake used ChatGPT and Figma, and his
5:08brother helped by doing a little
5:09engineering back end. But here's the
5:11part most people miss, and it's the real
5:13reason Umax spread. Blake didn't just
5:15build an AI face-rating app. He built a
5:18product that people wanted to share.
5:20Look at the design. It was a scorecard,
5:22and scorecards have a unique property.
5:24People naturally want to share them.
5:26Think about it. People post their exam
5:28scores, fitness stats, IQ tests, and
5:31Spotify Wrapped results. The moment you
5:33turn something personal into a number,
5:36people immediately want to compare it.
5:38Blake understood this. So, in his
5:40scorecard, you would get your current
5:41score and your potential score. Then
5:43below that, every trait gets broken down
5:46into its [music] own score. Jawline,
5:48skin quality, cheekbones, masculinity,
5:51and more. And that potential score is
5:53what made the product addictive because
5:55suddenly the app wasn't just saying,
5:57"Here's how attractive you are." It was
5:58saying, "Here's how much better you
6:00could become." The score gave people a
6:02reason to improve and come back. And the
6:04difference between the two gave people a
6:06story to share. If someone goes from a
6:0862 to a 74, they take a screenshot, post
6:11it on TikTok or Instagram, then their
6:13friends download Umax to see their own
6:15score. Those friends post their scores,
6:17and the cycle repeats. So, essentially,
6:19every screenshot became a free
6:21advertisement for Blake. That's why Umax
6:23[music] spread so quickly. And so, even
6:26the revenue started growing. In month
6:28one, they earned around $100,000.
6:30And by month two, it was $200,000. And
6:33since then, they've been doing half a
6:35million dollars regularly. Today, the
6:37app has crossed 10 million downloads
6:39only on Play Store. He had done it. The
6:42kid who had been sleeping in an attic
6:43and building broken apps just months
6:45earlier was suddenly running one of the
6:47fastest-growing consumer AI companies in
6:50the US. And that is exactly when someone
6:52tried to take all of it. Because the
6:54moment Umax proved this niche was worth
6:57money, the clones came. One of them,
6:59called Looksmash AI, copied the whole
7:01app top to bottom. And it started
7:03catching up fast. It matched Umax's
7:05download count in about 2 weeks. What
7:08Blake did next is the most aggressive
7:10move [music] in the entire story. It
7:11cost him more than the app had ever made
7:14him. And it is the reason he is still
7:15standing. Umax soon ended up outgrowing
7:18Looksmash AI by almost four times. So,
7:21Blake did not win because his app was
7:23better. He won because he got in front
7:25of the right people first. He won on
7:27distribution, not technology. And that
7:29is lesson one of the entire AI era.
7:32Write this one down. Technology is
7:33almost never the only moat. Anyone can
7:36copy the app. Anyone can use the same
7:37model. What people will struggle with
7:39achieving is the audience you already
7:41own. Which raises an obvious question.
7:43If the tech is so easy to copy, how do
7:46you build something that actually
7:47[music] lasts? Blake's answer was
7:49surprisingly simple. Stop chasing clever
7:51ideas. Instead, find a giant market that
7:54already exists and [music] make it 10
7:55times easier with AI. And soon enough,
7:57he found one. Calorie tracking. It's one
8:00of those habits that people want to
8:01stick to and almost nobody enjoys doing.
8:04Every meal means opening an app,
8:06searching for ingredients, estimating
8:08portions, and logging everything
8:10manually. It's tedious, repetitive, and
8:13[music] just annoying enough that most
8:15people eventually give up. Blake looked
8:17at that giant boring market and saw
8:19something different. He didn't need to
8:21create a new behavior. Millions of
8:23people already count calories. Millions
8:25more wish they did. He only needed to
8:27make the process dramatically easier.
8:30So, he decided to use the help of AI.
8:32[music] And so, he made Cal AI. You
8:34simply photograph your meal and the app
8:36estimates the calories and breaks down
8:38the protein, fats, [music] and carbs. If
8:40you look at it from a distance, it might
8:42seem like Blake was lucky with one app
8:44after another going viral. But, it
8:47wasn't just luck. It was a carefully
8:48executed playbook that he repeated over
8:51and over. And today, I'm going to teach
8:53you that playbook. One, the first lesson
8:56is about ideas. Most founders start with
8:58the technology [music] and ask, "What
9:00can I build with AI?" But, Blake starts
9:02from the opposite end. He asks, "What
9:05are people obsessing over already?" It
9:07could be dating, looks, status, [music]
9:09health, anything. These are problems
9:11that sit close to our deepest desires,
9:13>> [music]
9:13>> which means you never have to convince
9:15people to care. They already do. So, the
9:17first thing is to identify such a
9:19problem. Second, put a filter. It's
9:22called the three-word travel test. You
9:24should be able to pitch the app [music]
9:26in three words at a loud party, sober,
9:28and have the person turn to someone else
9:31within 10 seconds and say, [music]
9:32"Wait, did you hear about that?" For
9:34example, AI that texts girls for you or
9:37AI rates your looks or [music] even
9:39photo that counts calories. Those
9:41one-liners will get great word of mouth.
9:44If an idea needs a 5-minute explanation,
9:46[music] it is already dead. And thirdly,
9:48he focuses on timing. Blake's biggest
9:50insight is that billion-dollar
9:52opportunities often appear when two
9:54waves crash into each other, a cultural
9:57trend and a new technology. Dating
9:59anxiety collided with chat GPT [music]
10:01and Riz GPT was born. Looksmaxing
10:03collided with GPT 4 vision and Umax was
10:06born. So, how do you spot the wave
10:08first? [music] Blake watches three
10:09feeds. Feed one, a fresh TikTok account
10:12he trains with content from a group that
10:14is not him plus the niche subreddits and
10:17discords [music] where the obsessives
10:18live. Reading the repeated question in
10:20the comments because that repeated
10:22question is the unmet need. Feed two,
10:25the App Store top charts to see what
10:26[music] is climbing. And feed three, the
10:28release pages of OpenAI, Anthropic, and
10:31Google [music] because the week a new
10:32capability ships is the week a new
10:35category opens. And finally, he becomes
10:37the user. Blake did not survey
10:39looksmaxers. [music]
10:40He fed a burner TikTok nothing but
10:42looksmaxing content for 2 weeks and
10:44joined their world. [music] By the time
10:46he built his startups, he was one of
10:48them. So, he knew exactly which creators
10:50the community actually trusted. The
10:52second lesson is around executing fast.
10:55You're probably thinking you cannot
10:57code, but with AI you don't need to know
10:59how to code. All you need to do is sit
11:01[music] in front of a model and direct
11:03it. Blake treated AI like a junior
11:05engineer, gave it instructions screen by
11:08screen, and focused on directing rather
11:10than programming. For example, build me
11:12an app gets you nothing. But what if you
11:14said, "Build a screen with one button
11:17labeled
11:17>> [music]
11:17>> analyze photo. On tap, open the camera
11:20roll. Send the image to a vision model
11:23with this exact instruction [music] and
11:25show the result in a clean card." It's
11:27more specific and you get exactly what
11:29you envisioned. But more important is
11:31this tip, ship the ugly version. This is
11:34something that most founders struggle to
11:36do. Ris GPT launched with bugs, broken
11:39features, and even an exposed API key.
11:42Most people would have waited another
11:43month to polish it. Blake put it in
11:45front of users immediately because he
11:47believes the market teaches faster than
11:49perfection ever can. [music] And
11:51finally, the third lesson is that
11:52distribution is the actual product. UMAX
11:55exploded because its scorecards were
11:57designed to be screenshotted and posted.
11:59The product itself looked like a viral
12:01TikTok post. Blake [music] designed
12:02three product to be distributed. Blake
12:05found micro creators with attention
12:07instead of followers. The accounts Blake
12:09used had 50 to 100,000 followers, but
12:12millions of views per post. So, he paid
12:14small amounts to many creators and
12:16pitched them [music] the exact hook he
12:18wanted them to post. He is also
12:20relentless about it. He says, "You DM
12:22[music]
12:22about 100 creators, maybe 10 reply, and
12:25maybe three actually convert. And for
12:27the one creator he absolutely must
12:30have." He does not stop at one number DM
12:32on TikTok, no reply. DM on Instagram,
12:35join their Discord, message their
12:37manager, [music] make yourself
12:38unignorable because he has one very
12:40simple equation. If every thousand views
12:43makes you more money than it costs to
12:45buy those views, you can keep
12:47reinvesting and outrun everyone else.
12:49Now, there's a second playbook that's
12:51arguably even more important if you're
12:53getting [music] started today, which is
12:54the AI stack. The tools and technologies
12:57that Blake used to create these products
12:59that we wanted. [music]
13:00If you are a solo non-technical builder,
13:03here are the tools you can start with.
13:05Layer one is where the building happens.
13:07[music] So, in this case, tools like
13:08Claude Code and Cursor are the best to
13:11start with. Instead of writing software
13:13line by line, you simply describe what
13:15you want in plain English and the AI
13:17writes, explains, and fixes the code for
13:20you. It's essentially an engineer on
13:22demand. Layer two are the capability
13:24tools that power your product. UMAX
13:26needed image understanding, [music] and
13:28today the most powerful image generation
13:30tools are Nano Banana and ChatGPT's
13:32[music]
13:33image models. If you need a realistic
13:35voice, you can use Eleven Labs. And if
13:37you need AI avatars, the best tool to
13:39start with is HeyGen. Different
13:41capabilities require different tools.
13:43Layer three [music] is the store. Blake
13:45built iOS apps in Swift UI and
13:47distributed them through the App Store.
13:49[music] That's still a valid path, but
13:50it's not the only one. For many builders
13:52today, the fastest route is a simple web
13:55app with a payment button and one-click
13:57deployment. Finally, layer four is the
14:00marketing [music] and distribution. Now,
14:01note this because it's where most people
14:03fail. Finding an audience is just as
14:05important as building your product. For
14:08Blake, the solution was marketing
14:09through TikTok [music] creators. For
14:11others, it might be X, YouTube Shorts,
14:13Instagram Reels, newsletters,
14:16communities, or SEO. [music]
14:18And that's it. That's all you need to
14:19know within tools to start building like
14:22Blake. And that brings us to the final
14:23part of this video. The hard part
14:25[music] is no longer building, it's
14:27deciding what to build. Now, across the
14:29internet, people are already telling you
14:31exactly what [music] they want. You just
14:33have to pay attention. So, here are a
14:35few ideas we sourced from online
14:37communities that you could use to start
14:38building this weekend. The first is an
14:41AI data logger. If you type "track my
14:43workouts and mood," the app
14:45automatically creates forms, tables, and
14:47charts for you. Second is a subscription
14:49tracker. [music] It tracks every
14:51subscription you pay for and get
14:53step-by-step instructions on how to
14:55cancel the ones you no longer use. Third
14:57is a visitor sign-in app for small
14:59offices. Someone [music] scans a QR
15:01code, enters the name and company, and
15:04the system automatically logs the visit
15:06with a timestamp. [music]
15:07And fourth is an AI events concierge.
15:09Most people miss interesting events
15:11>> [music]
15:11>> simply because they never hear about
15:13them. This app scans local event
15:16listings and recommends the ones you're
15:17most likely to care about based on your
15:20interest. Notice the pattern. None of
15:22these ideas require breakthrough
15:23technology, need a big [music] team, or
15:25are trying to create demand. They are
15:27existing problems made dramatically
15:29easier with AI. So, by now, you have
15:32everything you need to start building.
15:34If you still need hand-holding for the
15:35tools, you can start with our Claude
15:37code masterclass. We literally walk you
15:39through the tool step-by-step and build
15:41real projects alongside you. It is the
15:44fastest way to go from watching this to
15:46having something live tonight. So, close
15:49this video, open your laptop, and go
15:51make something. And to not miss out on
15:53more such videos on stories of AI indie
15:56builders like this, subscribe to our
15:58channel. I'll see you in the next one.

About this transcript

This page contains the full transcript of ChatGPT Made This Broke Kid $1 Million a Month in 18 Months by Vaibhav Sisinty, generated from the public captions YouTube serves with the video. The transcript has 3,032 words across 464 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.