---
title: 'How to Build a $20 Million App Idea in 18 Minutes Using AI'
source: 'https://youtube.com/watch?v=SI8T1yvmoRs'
video_id: 'SI8T1yvmoRs'
date: 2026-07-12
duration_sec: 916
---

# How to Build a $20 Million App Idea in 18 Minutes Using AI

> Source: [How to Build a $20 Million App Idea in 18 Minutes Using AI](https://youtube.com/watch?v=SI8T1yvmoRs)

## Summary

This video demonstrates how to build a profitable app business using AI, specifically by reverse-engineering the success of Cal AI, a photo-based calorie tracking app that generates $20 million annually. The host uses Manus AI to research user feedback, generate app ideas, and build a functional app called Pantry Chef AI in under 20 minutes. Additionally, it covers a second income stream via Amazon KDP, focusing on low-competition niches like postpartum recovery journals.

### Key Points

- **Cal AI's Success** [00:00] — An 18-year-old built Cal AI, a photo-based calorie tracking app, generating $2 million in one month and $20 million annually by solving the friction of manual food logging.
- **AI-Powered App Building** [00:40] — Using Manus 1.6 Max, an AI agent, the host builds an entire app without coding. The key is research first: analyzing over 500,000 Cal AI reviews to identify user likes, complaints, and missing features.
- **User Preferences** [02:06] — Users prioritize speed over accuracy; small errors don't deter them. Major complaints include accuracy issues (e.g., labeling an apple as tikka masala), billing problems, and performance freezes.
- **User Desired Features** [02:58] — Users want editable starting weight, history of eaten foods, portion control, full nutrition details, lock screen access, and transparent pricing before signup.
- **Five App Ideas Generated** [03:37] — Manus AI produces five app ideas: Form Check AI, Pantry Chef AI, SleepSync AI, Grocery Run AI, and Mindrep AI. The host chooses Pantry Chef AI for its universal problem-solving.
- **Pantry Chef AI Build** [05:05] — The AI builds the app's front end, backend, and core features. It scans fridge photos, identifies ingredients, and suggests recipes with macro tracking. The app includes dark mode, calorie ring, and grocery list integration.
- **Adding Features and Design** [07:30] — The host adds a 'save favorites' feature and upgrades the design to a premium dark theme with orange accents, making it look like a paid product.
- **Deployment** [08:42] — For Android, Manus provides an APK for direct install or Play Store upload. For iOS, the code can be submitted to Apple. The entire build took under an hour.
- **Amazon KDP Strategy** [09:21] — The second income stream is low-content books (journals) on Amazon KDP. The key is researching market demand to find low-competition niches.
- **Niche Selection** [10:19] — Manus analyzes KDP and identifies postpartum weight loss and recovery as a high-demand, low-competition niche (fewer than 1,000 competitors vs. 30,000 for generic fitness journals).
- **Journal Creation** [11:42] — Using Manus's design mode, the host creates a 90-day postpartum journal with cover, daily trackers, weekly reflections, and a certificate of completion. The design can be customized with AI-generated art.
- **Design Customization** [13:11] — The design view allows real-time editing of text, colors, and images. The host replaces cover art with a custom watercolor illustration of a mother and baby.

### Conclusion

The video presents two viable income strategies: building AI-powered apps and publishing low-content books on Amazon KDP. Both rely on thorough market research to identify real user needs, and the host emphasizes focusing on one strategy to achieve mastery.

## Transcript

Last month, the app generated just over $2 million. >> This 18-year-old just made $2 million in one month from a photo-based app he built. That's more than most people [music] make in 5 years. In one month.
I'm reverse engineering his entire app business step by [music] step. Then showing you a second strategy on Amazon. Two ways to win. Pick one or run both. Let's go. That's Zach Yadagari. His app is called Cal [music] AI and it's doing
$20 million a year. He solved one massive problem. People hate tracking their food manually. So, he turned a 10-minute task into a two-cond photo. You snap a picture of your meal. The AI calculates the calories. Done. That's
the entire business, and it prints money. Now, you're probably thinking, "That's cool, but I'm not a coder. I don't have $100,000 to hire developers. I'm about to prove you wrong." This is Manis 1.6 Max. It's an AI agent that
builds the entire app for you. But before we hit build, we have to do what 99% of people skip. Research. Most people build things nobody wants. We're build the solution. Let me show you how. All right, I'm opening up Manis AI. I
exactly what I'm doing in real time. Now, look at these toggles. Light, regular, and max. For deep research that actually finishes the job, you go max.
This architecture gives you 30% more persistence. It doesn't quit halfway through. It keeps digging until it's done. All right. So, here's what I'm telling it. Research Cal AI app reviews. Find what users love, their top
complaints, and features they're requesting but missing. Now, I just let it run. Watch what happens. I am doing nothing while the AI works in the background. It reads reviews, checks competitors, and pulls data from the App
Store, Google Play, Reddit, and other sources. The scan finishes after reviewing more than 500,000 real user reviews for Calai. Here is what users like most. People love how easy the app feels to use. One user said, "I point,
click, and it logs it." This is for lazy people who want results. This shows a clear pattern. People choose speed over accuracy. If logging takes only a few seconds, small mistakes do not stop them from using the app. Now, let's talk
about the problems. Some users report serious accuracy issues. In one test, the app labeled an apple as tikka masala. Errors like this damage trust and make users question the system. Billing is another major issue. Some
users canled a 3-day trial, but still got charged for a full year. [music] When they contacted support, no one replied. Performance issues also appear often. The app freezes and users say they cannot edit their entries. Here's
the surprising part. Even with all these problems, the app makes around $20 million per year. This shows one clear truth. Solving the main friction point matters more than [music] fixing every issue. Now, let's look at what users
want but do not get. Users want to edit their starting weight because the first entry affects all progress. They want to see what they already [music] ate, not only what is left. They want portion control, so half a meal logs as half.
They want full nutrition details like sugar and fiber. They want faster access from the lock screen. They also want clear pricing shown before signup. Now we understand the market clearly. We see what works, what fails, and what people
still pay for. This insight comes from real user data, not guesses. Now let's go deeper. Based on this research, generate five app ideas that solve similar friction problems in health and fitness. For each idea, include one, the
specific problem it solves, two, how the AI would work, and three, why it scales to millions of users. Include concrete examples. Look at this. Manis AI produced five app ideas, each solving a real daily
problem. One is form Check AI, where you place your phone during a workout and the AI corrects your form in real time. Another is Pantry Chef AI, where you already have while matching your diet goals. There's also SleepSync AI, which
tracks your habits and warns you before you hurt your sleep, like telling you when to stop caffeine. Grocery Run AI helps you set a diet goal and budget, then builds meals using food on sale near you. The last one is Mindrep AI, a
ideas are solid, but I am choosing pantry chef AI. It fixes the exact gap we saw with Cal AI. Everyone owns a fridge and everyone asks the same question. What do I cook right now? This is a universal problem with a fast AI
answer. So, this is the one we build. Now, watch this. I am not coding or designing anything. I am describing what I want. I tell the AI to build a mobile app for Android called Pantry Chef AI. Users take a photo of their fridge. The
AI identifies the ingredients, checks their remaining calories, and suggests a few meals that fit. If something is missing, it shows it. The design stays clean, and the app is ready to test right away. That's it. Let's see what it
All right, the AI completed the full build process. It generated the front end, set up backend logic, and connected all core features. Once the build
finished, I scanned the QR code. Look at this. Pantry Chef is running live on my phone. The logo was generated by the AI using Nano Banana Pro, and I like how it looks. I did not design any of this. Manis built the entire interface. It
comes with dark mode, a clear calorie progress ring, and automatic breakdowns for protein, carbs, and fats. Remember the Cal AI research? Users liked photo tracking and they wanted help planning meals using food they already had. This
app solves both. Watch this. I tap scan ingredients and take a photo of my fridge. The app processes the image and finds 39 ingredients. It identified everything inside. As I scroll, you can see all 39 ingredients listed clearly.
Each one recognized and shown separately. If anything is missing, I tap add missing ingredient and enter it manually. Next, I tap find recipes. It does not show generic ideas. It generates options based on my
ingredients. Take this example, a high protein yogurt and berry power bowl. It shows a 95% match because I already have almost everything. Using the ingredients I own, it gives me step-by-step instructions to make the recipe. Here is
the part I really [music] like. It knows I am missing one ingredient. I tap the yellow cart icon, and it adds only that item to my grocery list. The connection between the fridge, the recipes, and the store is smooth. This is not a simple
mockup. In settings, I edit my daily macro goals. The app saves my preferences in the back end and adjusts [music] results based on my targets. Now, let's bring this back to the business side. Cow AI makes around $20
million a year by removing friction from tracking. This app does the same thing and also solves the biggest user complaints. Detailed macros and cooking based on existing inventory. I did not code this. I did not design this. I
described a problem. The AI built a solution. And 18 minutes later, it is a working business on my phone. But let's make it better. Because if I were actually using this, there's one thing I'd want. Saving favorites. [music] When
you find a recipe you like, you should keep it. Simple. So, I tell the AI to add a save favorite recipes feature. Heart icon on every recipe card. Tap it once, it's saved. No friction. [music] Watch this. It rebuilds the app with the
new feature. And done. Now, the hearts are there. I tap one, the recipe saved. I open the favorites tab and everything I liked is sitting there. Over time, this turns into your own personal cookbook. Now, one more upgrade. This is
about perception. functionality is there. Now it needs to look premium. So I tell it to clean up the design. Dark theme, orange accents, minimal layout, clear visuals. Nothing fancy, just sharp. It rebuilds again. And look at
this. The same app, but now it feels expensive. Clean dark mode, polished progress rings, clear macro visuals. This looks like a product people would happily pay for. Same product, same features, better design. All done with a
single prompt. So, where do we go from here? If you're on Android, Manis literally hands you the APK file. You can install it on your phone right now or upload it to the Google Play Store and start getting downloads immediately.
Your business is live. On iOS, you just take the finished code, upload it to your Apple developer account, and submit it for review. The hard part, building the app, is already done. And that is the key difference. Most no code tools
just give you a web demo. Manis gives you a deployable product that you can Lambo. He spent 10 years learning to code. You just closed that gap in under
an hour. All right, app strategy covered. Now, let's talk about the second income stream. Maybe you don't want to build apps. Maybe you want something simpler that pays you faster. That brings us to Amazon KDP, [music]
specifically lowcontent books like journals. Now, you've seen the screenshots. people pulling in 5K, 10K a month on autopilot. But for every success story, there are 10,000 people making exactly 0. Why? Because they're
guessing. They made something they liked, not what the market actually wanted. The solution is simple. Find out what people are already buying, then create that. We're going to use the same research process we used for the app.
I'm starting a new task in Manis and setting it to 1.6 max for deep research. Here is exactly what I'm asking. Research Amazon KDP best-selling journals in health and wellness. Find the top selling journal types, common
features that convert, niches with high demand but low competition, and best price points. Focus on weight loss and fitness journals. Let's see what the market wants.
All right. Manis analyzed the KDP market. It filtered out the crowded markets and found the gaps, areas with high demand but low competition. The short list included senior fitness, specific diets, and women's recovery. I
compared them all, but I focused on one key factor, urgency. That is why postpartum weight loss and recovery is the winner. The first 90 days after birth are critical. New moms want safe recovery, habit tracking, and structure
to feel stable again. A generic fitness journal just doesn't work for this phase. They need something built specifically for them. [music] And look at the numbers. The AI found fewer than a thousand competing products in this
niche. Compare that to generic fitness journals which have over 30,000 competitors. That is a massive difference. Monest recommends a 90-day transformation format to match the fourth trimester. It also suggests
pricing it at $14.99. That is a premium price, but because we're solving an emotional problem, not just selling a notebook, people will happily pay it. New mothers are actively searching for guidance. Yet almost no
products address their specific needs. That gap is where the profit is. The research just handed us the winning idea. Now, let's create it. I'm starting a new task in Manis and clicking on design. This activates the latest Nano
Banana Pro model, which is their advanced visual engine. I'm telling it, create a complete 90-day postpartum weight loss and recovery journal for Amazon KDP. Now, for a real printed book, I would normally generate about
108 pages, but for this video, I'm going to set it to 21 pages just so I can show you the variety quickly. Let's hit generate. And look at this. It's creating the entire journal from scratch. First, the
cover design. clean, motivating with a soft, feminine aesthetic. Then look at the interior. It starts with an about me page where the mom can fill in her name, the baby's name, and the birth date. Next, it built a 90-day goals section.
It has a starting point for current stats and a dedicated placeholder to attach a before picture so they can visualize their transformation. It's not just blank lines, either. We have daily tracker pages, weekly reflection
sections, and full end of month reviews. It even added niche specific value, nursing friendly workouts, recovery milestones, and motivational quotes throughout. [music] And at the very end, a certificate of completion celebrates
finishing the 90-day journey. Honestly, I am speechless. I know how hard it is to design layouts like this manually. It usually takes days of formatting, [music] layout, design, typography, page structure, visual consistency across 108
This is the new design view. It's not just a preview window. It's your command center. Most AI tools just spit out an image and say, "Good luck." Here, I have
full control. If I want to change colors, move things around, or fix the text, I do it right here. Let me show you an example. I want to change the text on this cover. I just click edit text, type what I want, and it changes
instantly. I can also upscale the quality or remove backgrounds with one click. But look at this cover art. The gold lines are nice, but for a premium journal, I want something more emotional. I want a custom piece of art.
[music] So, watch this. I select the mark tool and highlight the image area right here on the canvas. It adds it to the prompt box. And I'm going to tell the AI exactly what I want. Soft watercolor illustration of a mother
gently holding a newborn baby. Ethereal style. Pastel pinks and gold accents. Dreamy and emotional. High resolution. Hit generate. And boom. Look at that
difference. It instantly took the design from basic to best seller quality. It's the same process for anything. Colors, fonts, images. Once you are happy, you just download it to your computer. This file is ready to upload to Amazon KDP
right now. Now, if you are new to Amazon KDP and don't know how to set up an account or upload files, don't worry. I have a full free tutorial linked in the description. It covers everything you need to start. So, I showed you two
game-changing strategies today. One, the tech startup building apps. Two, the digital publisher Amazon KDP. Comment below which one you like most. I will make more videos based on what you vote for, but listen to me closely. This is
the most important part. Pick one. The man who chases two rabbits catches neither. Do not try to do both of these at the same time. Just pick the strategy that fits you best, stick to it, and master it. That is how you win. God
bless and I'll see you in the next