RizzGPT: How a Broke 20-Something Built a Million-Dollar AI Empire From His Brother’s Attic

Picture this. You’re sleeping on a mattress in your older brother’s attic. You can’t afford groceries, so he’s Venmo-ing you money just so you don’t go hungry. Eighteen months later, you’re pulling in close to a million dollars a month and running one of the fastest-growing AI app companies in the country.

That’s not a movie pitch. That’s the real story of a young founder named Blake, and it starts with an app most people wrote off as a joke: RizzGPT. If you’ve never heard of it, stick with me, because this is one of those stories that sounds made up until you look at the numbers.

I’ve spent years writing about startups and side hustles, and honestly, this one made me put my coffee down. Not because Blake invented some jaw-dropping technology — he didn’t. He’s not even an engineer. What he figured out is something way more valuable, and way more copyable for the rest of us: how to spot a wave before everyone else jumps on it.

Let’s get into it.

What Exactly Is RizzGPT?

RizzGPT started as a stupidly simple idea. Blake’s college roommate was a genuinely nice guy who kept striking out on dating apps — not because he wasn’t likeable, but because he froze up the second a match replied. He just didn’t know what to text back.

Blake watched this happen over and over, and he had a realization that most people would’ve missed. His roommate didn’t need dating advice. He needed someone — or something — to just write the message for him.

So that’s what RizzGPT did. You’d screenshot your conversation, the app would read the text off the image, feed it into ChatGPT, and spit back a handful of flirty reply options. There was even a “spice meter” so you could dial the boldness up or down depending on how bold you were feeling.

It sounds almost too basic to work. And yet, it worked in a big way.

The Rocky, Chaotic Launch

Here’s the part that surprised me most. RizzGPT’s launch was, by any normal engineering standard, a mess:

• Notifications barely functioned
• There were basically no reviews
• Blake accidentally exposed his API key to the public

Most founders would’ve pulled the app and quietly patched everything before showing it to a single stranger. Blake did the opposite. He shipped it broken, bugs and all, because he understood something a lot of “serious” developers overlook: on the internet, getting in front of people matters more than being polished.

The Growth Hack That Actually Worked

Instead of throwing money at Facebook or Google ads, Blake went hunting in a weird little corner of TikTok — anonymous accounts posting pickup lines and dating advice, racking up millions of views despite having barely any real following.

These accounts were basically distribution machines nobody was paying attention to yet. Blake paid two of them $50 each. That’s it. A hundred bucks total.

The creators slipped RizzGPT into their slideshows so naturally it didn’t feel like an ad — it felt like the exact answer their audience had already been asking for in the comments. And then the internet did what the internet does.

Within about a week, the app crossed 200,000 downloads. What started as a favor for a lonely roommate was suddenly making around $80,000 a month. Years later, it’s still pulling close to $200,000 monthly, now rebranded as Plug AI after another company pushed back on the name.

“You don’t need to invent new technology. You just need to stand where a cultural trend and a technological shift collide.”

That one line pretty much sums up the entire philosophy behind everything Blake built next.

From RizzGPT to a Full-Blown AI App Empire

RizzGPT was never meant to be the final destination. It was proof of concept — proof that Blake could spot a cultural wave before anyone else and build exactly what it needed. And once he had that confidence, he went looking for a bigger wave.

App Two: UMAX, the Face-Scoring App That Broke the Internet

If you’ve spent five minutes on TikTok, you’ve probably seen the “looksmaxing” trend — young guys comparing jawlines, rating each other’s faces, asking strangers to be brutally honest about their appearance.

At the exact same time, OpenAI dropped GPT-4 Vision, which meant AI could suddenly look at a photo and actually understand it. Before that moment, a face-rating app would’ve needed custom computer vision models, a real engineering team, and months of work. Suddenly, one API call could do most of the heavy lifting.

Blake saw the same pattern that made RizzGPT explode: a massive cultural obsession colliding with brand-new tech, and almost nobody standing at that exact intersection.

So he built UMAX. Upload a selfie, get scored across 20+ traits, and get told what’s actually fixable — skin, hairstyle, grooming, body fat — not your bone structure.

Why UMAX Spread Like Wildfire

This is the part every marketer should be taking notes on. UMAX didn’t just rate your face. It handed you a scorecard, and scorecards are inherently shareable. Think about it — people post their exam results, their Spotify Wrapped, their fitness stats. The second something personal becomes a number, people want to compare it.

UMAX showed you your current score and your “potential” score side by side. That gap became the hook. Go from a 62 to a 74 and you screenshot it, post it, and suddenly your friends are downloading the app just to see their own number.

Every single screenshot became free advertising. Revenue went from $100,000 in month one, to $200,000 in month two, to steady half-a-million-dollar months not long after. Today it’s crossed 10 million downloads on Google Play alone.

The Copycat War

Success invites clones, and UMAX got hit hard. An app called Looksmax AI copied it top to bottom and matched its download numbers within two weeks. Blake’s response was aggressive — he outspent and out-marketed the copycat until UMAX was pulling in roughly four times the downloads.

He didn’t win because his app was technically better. He won because he had the audience already. That’s lesson one, and it’s worth writing on a sticky note: technology can be copied by anyone. An audience takes real work to build, and that’s the actual moat.

App Three: Cal AI and the Power of Boring Markets

After the copycat scramble, Blake changed his approach. Instead of chasing another clever idea, he went after a giant, boring, already-proven market: calorie tracking.

Nobody enjoys logging every meal manually. It’s tedious. Most people quit within a few weeks. But millions of people already do it, and millions more wish they did. Blake didn’t need to invent a new habit — he just needed to remove the friction.

Cal AI lets you photograph your meal, and the app estimates calories plus a breakdown of protein, fat, and carbs. Simple, boring, and genuinely useful — which turned out to be exactly the point.

The Real Playbook Behind It All

Here’s the thing — none of this was luck. It’s tempting to look at RizzGPT, UMAX, and Cal AI and think Blake just got lucky three times in a row. But when you zoom out, there’s a clear, repeatable system underneath it all. Let’s break it down.

Step 1: Find the Obsession, Not the Technology

Most builders start by asking, “what can I build with AI?” Blake starts from the opposite direction and asks, “what are people already obsessing over?” Dating, appearance, health, status — these sit close to people’s deepest desires, so you never have to convince anyone to care. They already do.

Step 2: The Three-Word Test

Once you’ve got an idea, run it through a filter Blake calls the three-word test. You should be able to describe your app in three words at a loud party and have the listener immediately turn to someone else and repeat it.

• “AI that texts girls for you” — this became RizzGPT
• “AI rates your looks” — this became UMAX
• “Photo that counts calories” — this became Cal AI

If your idea needs a five-minute explanation before anyone gets it, it’s already dead in the water.

Step 3: Watch for Colliding Waves

Blake’s biggest insight is that the really big opportunities show up when a cultural trend crashes into a new technology at the same time. Dating anxiety plus ChatGPT gave us RizzGPT. Looksmaxing plus GPT-4 Vision gave us UMAX.

To actually spot these waves before they’re obvious, he watches three feeds constantly:

• A fresh TikTok algorithm trained on niche content, plus the subreddits and Discord servers where the obsessives hang out — reading the repeated question in the comments, because that repeated question is the unmet need.
• App Store top charts to see what’s actively climbing right now.
• Release pages from OpenAI, Anthropic, and Google, because the week a new capability ships is the week a brand-new category quietly opens up.

He even went as far as feeding a burner TikTok account nothing but looksmaxing content for two weeks so he could genuinely understand that community before building anything for them.

Building Fast Without Knowing How to Code

Here’s the part that should get your attention if you’ve ever told yourself “I can’t build an app, I’m not technical.” Blake isn’t an engineer. He never built an app before RizzGPT. He just learned how to direct AI like it was a junior developer on his team.

The trick isn’t typing “build me an app.” That gets you nothing useful. Instead, you break it down screen by screen:

“Build a screen with one button labeled ‘analyze photo.’ On tap, open the camera roll, send the image to a vision model with this exact instruction, and show the result in a clean card.”

Specificity is everything. And once it’s built — ship it ugly. RizzGPT launched with bugs and an exposed API key, and Blake put it in front of real users anyway, because the market teaches you faster than months of internal polishing ever could.

Distribution Is the Actual Product

If there’s one takeaway that matters more than any coding tip, it’s this: distribution isn’t something you bolt on after building your app. It’s part of the product itself.

UMAX’s scorecards were designed from day one to be screenshotted and shared. That’s not an accident — that’s product design working hand-in-hand with marketing.

On the outreach side, Blake targeted micro-creators — accounts with 50,000 to 100,000 followers but millions of views per post, because engagement matters more than raw follower count. His approach to outreach was relentless:

• DM about 100 creators
• Expect maybe 10 replies
• Expect roughly 3 actual conversions
• For the one creator you absolutely need, don’t stop at one DM — try Instagram, join their Discord, message their manager

His logic is simple: if every thousand views makes you more money than it costs to buy those views, you keep reinvesting and you outrun everyone else who’s still waiting for the “perfect” launch.

The No-Code AI Tool Stack

If you’re a solo builder with zero engineering background, here’s roughly the stack that made all of this possible:

Layer	Purpose	Example Tools
Building	Writing and fixing code from plain English instructions	Claude Code, Cursor
Capabilities	Image, voice, and avatar generation powering the product	Nano Banana, ChatGPT image models, ElevenLabs
Distribution store	Getting the product in front of users	App Store (Swift UI), or a simple web app with payments
Marketing	Getting people to actually find the product	TikTok creators, YouTube Shorts, Instagram Reels, SEO

The point isn’t that you need every single one of these tools. It’s that you need one from each layer, and you need to move fast between them instead of getting stuck perfecting layer one.

5 App Ideas You Could Realistically Build This Weekend

Here’s the good news — you don’t need a brand-new billion-dollar idea. You just need an existing, boring problem made dramatically easier with AI. A few ideas sourced from online communities:

• An AI data logger — type “track my workouts and mood” and the app auto-generates forms, tables, and charts.
• A subscription tracker — tracks every recurring charge and gives step-by-step cancellation instructions for the ones you don’t use.
• A visitor sign-in app for small offices — scan a QR code, enter your name, get logged with a timestamp automatically.
• An AI events concierge — scans local listings and recommends events based on your actual interests.
• A face or style-scoring app in the same family as UMAX, but applied to a different niche audience.

None of these need breakthrough technology or a big team. They just need someone willing to actually ship them.

Pro Tips If You Want to Copy This Playbook

• Start with the audience, not the app. Find where a group of obsessed people already hangs out online before you write a single line of code.
• Test your idea out loud first. If you can’t explain it in three words at a party, simplify it until you can.
• Ship before you feel ready. A broken app with real users teaches you more in a week than a perfect app sitting on your hard drive for a month.
• Design for screenshots. If your product produces a number, a score, or a result, make it visually satisfying to share.
• Pay for reach, not just ads. Micro-creators with high engagement are often cheaper and more effective than big-name influencers.
• Watch AI release notes like a hawk. New model capabilities open brand-new app categories almost every month.

Common Mistakes People Make Trying to Copy This

• Over-engineering the first version. Spending months polishing before anyone’s even seen the product is the fastest way to lose momentum.
• Chasing the tech instead of the audience. Building something cool that nobody was actually asking for rarely goes anywhere.
• Ignoring distribution until launch day. By then it’s too late — distribution needs to be baked into the product itself, like UMAX’s scorecards.
• Giving up after one round of outreach. Remember, out of 100 creator DMs, maybe 10 reply and 3 convert. That’s normal, not failure.
• Copying the exact idea instead of the underlying pattern. The next winning app probably isn’t another face-scoring tool — it’s whatever obsession-plus-technology combo nobody’s built yet.

A Quick Beginner’s Guide to Getting Started

If all of this sounds exciting but a little overwhelming, here’s a simplified path to follow:

1. Pick a niche you’re already curious about — dating, fitness, productivity, whatever pulls you in naturally.
2. Spend two weeks lurking in the subreddits, Discords, and TikTok comment sections that community lives in.
3. Write down the one question people ask over and over. That’s your unmet need.
4. Sketch the simplest possible version of a solution — one screen, one button, one clear action.
5. Use an AI coding tool like Claude Code or Cursor and describe each screen in plain, specific English.
6. Ship it, bugs and all, to a small group of real users.
7. Find five to ten micro-creators in that niche and offer them a small flat fee to try it out.
8. Watch the numbers, iterate fast, and reinvest anything that’s working.

None of these steps require a computer science degree. They require patience, curiosity, and a willingness to look a little unpolished in public for a while.

FAQs About RizzGPT and Blake’s Playbook

Is RizzGPT still around today?

Technically, yes, but it now goes by a different name — Plug AI — after a legal dispute over the original branding. The core idea is still the same: AI-generated reply suggestions for dating app conversations.

Do I need to know how to code to build something like RizzGPT?

No, and that’s really the whole point of this story. Blake wasn’t an engineer when he built his first app. With tools like Claude Code and Cursor, you describe what you want in plain English and the AI handles most of the actual coding.

How much money did RizzGPT actually make?

Within about a week of launch, it crossed 200,000 downloads and was generating roughly $80,000 a month. It’s reportedly still bringing in close to $200,000 monthly under its new name.

What made UMAX grow faster than RizzGPT?

The scorecard format. Turning your face into a shareable number gave people a natural reason to post their results, which turned every user into a free marketing channel.

Is this kind of app-building strategy sustainable, or just a trend?

The specific apps might fade as trends shift, but the underlying method — watching for a cultural obsession colliding with a new technology — isn’t going anywhere. New AI capabilities keep shipping, and new categories keep opening up as a result.

What’s the biggest lesson from the whole RizzGPT story?

Distribution beats perfection. A rough, buggy app that reaches the right audience will always outperform a polished app that nobody ever sees.

Final Thoughts

The RizzGPT story isn’t really about dating apps, or face scores, or calorie counting. It’s about a way of looking at the world that most of us have been trained out of — noticing what people are already obsessing over, and asking what happens if you make that obsession ten times easier.

You don’t need a computer science degree, a funded startup, or some secret insider knowledge of AI to try this yourself. You need curiosity about a community, the discipline to actually ship something imperfect, and enough hustle to DM a hundred creators when ninety of them won’t reply.

So here’s my honest advice: close this tab, open a notes app, and write down three things people around you obsess over. Then go lurk in that world for a week. Somewhere in those obsessions is probably your own version of RizzGPT, just waiting for you to notice it.

Want more real breakdowns of how indie builders are using AI to launch profitable apps? Check out our full library of founder stories here, or dive deeper into the tools mentioned in this piece over on Anthropic’s official documentation.

NVIDIA Story: How a Chip Maker 30 Days from Bankruptcy Became Worth $4 Trillion

BTech in 2026: The Honest Truth Nobody Tells You Before You Enroll

Startups Indian Keep Running at a Loss — The Brutal Truth Nobody Tells You