AI with Kyle Daily Update 126
Today in AI: $100M app refresh opportunity
Watch this video on our dedicated watch page
Better viewing experience with related videos and full-screen player
What’s happening in the world of AI:
Highlights
Greg Eisenberg dropped a playbook that's been doing the rounds on Twitter. I've been elaborating on it because I think there's a massive opportunity here that very few people are rushing into yet. Here’s the original tweet article.
The basic idea: find apps that people pay for but absolutely hate, then rebuild them using Claude Code in an afternoon.
I did exactly this last week out of spite! But there’s a bigger opportunity here. Here's the full guide.
📱 The Opportunity
The App Store and Google Play are stuffed with predatory apps charging subscriptions for basic functionality. They have millions of downloads, terrible reviews, and they're making serious money.
Example: I needed an app to copy a URL from my phone to my Android TV. Every remote control app on the store follows the same gross model - free to download, then $3-5/month to use basic features. Intrusive ads. One-star reviews everywhere: "I am unhappy," "Way too many ads," "You have to pay to turn on your TV."
One of these apps has 50 million downloads and 650,000 reviews.
I refused to pay. Out of spite, I opened Claude Code, gave it screenshots of these apps, and said "build me this." 15 minutes later I had a working app. Another 15 minutes to get it through Xcode and onto my phone via TestFlight.
Half an hour total. That's the opportunity.
I checked (sources below) and some of these remote control apps are making $100,000 a month. Ridiculous.
Let’s look at how to roll this strategy.
🔍 Step 1: Hunt for Hated Utilities
Find apps making money that people despise. You're looking for:
Monthly revenue of $50,000 - $200,000.
App Store rating below 4 stars.
Recent reviews full of one-star complaints about pricing, ads, or broken functionality.
Tools for research:
Sensor Tower - The gold standard. Used by enterprises. Most accurate data. Not cheap (hundreds per month). Use this if you're serious.
AppArc (appark.ai) - Free. Shows estimated revenue and downloads. Data isn't as accurate but points you in the right direction.
Fox Data (platform.foxdata.com) - Also free. Similar functionality. Good for validation.
My recommendation - start with free versions. You don't need exact numbers. You need direction, not acceleration. You just need to find niches where apps are making money but users hate them.
🎯 Step 2: Identify Why They're Hated
Look through the one-star reviews. Common patterns:
Technical debt - App hasn't been updated in years, login loops, things don't work anymore.
Hostile UX - Paywalling basic features, intrusive ads, dark patterns.
Missing obvious features - Single-dev apps that never added what users kept asking for.
Use Claude to help here. Throw in all the app descriptions and reviews from competitors and ask: "What do people hate most? What's missing? How could this be 10x better?"
🏆 Step 3: Pick Boring Categories
Don't build the next Instagram. Don't chase AI photo generators. Build boring shit.
Target: utilities, productivity tools, professional apps. Things people need daily but that aren't sexy.
Greg's example: Alarmy - a loud alarm clock app. That's it. It's louder than the default. Half a million downloads per month. Half a million dollars per month. Again that’s it. Build boring.
Key rules:
Don't build on APIs you don't control. For instance, avoid connecting to social media APIs - that adds risk and ongoing costs.
Copy what works and make it 10x better. You're not inventing anything. You're executing better on proven demand.
🛠️ Step 4: Build with Claude Code
Before opening Claude Code, plan first. Duh. Talk to Claude or ChatGPT and say:
"I'm rebuilding this app. Here are screenshots. Here are competitor descriptions. Here are the reviews showing what people hate. I want to build for iPhone/Android. This is my first app - what do I need to know?"
Open Claude Code and build. Give it the screenshots, describe the functionality, tell it the target platform. It'll figure out the languages and frameworks based off your plan.
🚀 Step 5: Publish
Getting from built app to App Store used to be the painful part. That's changing. Slowly, but it’s changing.
There are a couple of tools that can help you publish.
Manus just launched app publishing. Build your app, click publish, it handles the packaging and TestFlight integration. It even helps you plan testing.
Direct through Xcode/Android Studio - More friction but totally doable. Claude Code can walk you through every step.
FYI Developer account costs: Apple is $99/year. Google Play is $25 one-time. You can test and play around with free accounts initially but there will be a cost to publish.
💰 Step 6: Price It Better
Most of these predatory apps charge subscriptions for features that have zero ongoing cost. A TV remote doesn't need monthly payments - there's no server, no API calls. They are just being greedy bastards. Be better!
Options:
One-time fee - $5 lifetime instead of $3/month. People hate subscriptions.
Donation model - App is free, "support my work" option.
Freemium - Basic free, premium one-time for advanced features. Keep it reasonable.
Only charge subscriptions if you have actual ongoing costs (like AI API calls). If you don’t have ongoing costs maybe charge a one-time fee? Your call of course!
📣 Step 7: Marketing (The Hard Part)
Building is now easy. Marketing is the differentiator.
This is too big to cover in one newsletter, but I've got a free 10-week course on my site that covers: finding your niche, building in public, getting your first customers, bootstrapping sales, and scaling.
→ Free AI Business Course (free)
Remember we're not trying to build billion-dollar unicorns. We're trying to build something that generates $10,000-50,000/month. That's absolutely life-changing money for an individual. And that's the kind of revenue big companies won't bother competing for. It’s chump change.
Member Question: "You are overhyping vibe coding, in my opinion. What about maintenance?"
Kyle's response: Fair point. The specific apps we're talking about - simple utilities - don't need much maintenance. The Android TV interface hasn't changed since 2020. If something does break and you're making $10-30K/month, you can either hire someone or go back to Claude Code and fix it.
Just don’t let fears of future failure get in the way of starting.
Also, re: overhype…if you're a software engineer, vibe coding is an absolute nightmare and you have every right to hate it. If you're an entrepreneur with ideas, it's a godsend. It lets you build an MVP, get it to market, and find out if anyone cares - before spending months and thousands of dollars.
If anything it’s under-hyped for entrepreneurs!
Member Question: "Can Claude Code created code be moved into Manus?"
Kyle's response: Yes - connect through GitHub. Make sure your Claude Code project is in a GitHub repository. Then connect Manus to the same repo. Manus will pull the code and you can use their publish-to-App-Store feature. Moving files manually would be a mess, but GitHub makes it clean.
Member Question: "Can Claude Code help for an app that pulls data from other websites and public data?"
Kyle's response: Yes. Either have Claude Code build a scraper for you, or connect it to a scraper service like Apify. Apify has pre-built scrapers for Instagram, Amazon, Google Maps, TikTok - hundreds of them. Building your own scraper is a pain because you need to handle detection avoidance, IP rotation, etc. Apify handles all that. Costs are reasonable - about $2.30 per 1,000 Instagram comments, for example.
Member Question: "Is it good to make a religious app? Is it too sensitive?"
Kyle's response: Absolutely works. Marcin did a TikTok about Bible Note - a faith-based app doing $50K/month. Daily Bible quotes, journaling prompts, simple stuff. The Bible is public domain. Religious apps monetise habits, belief, and commitment - some of the strongest forces on earth! I bet there’s a HUGE market here.
Member Question: "How do you price an app?"
Kyle's response: Complex question, but let’s give it a shot: if there are no ongoing costs, don't charge a subscription - that's a bit gross in my opinion. Instead charge a one-time fee. If you have ongoing costs (AI API calls, servers), subscription makes sense and are fair.
When in doubt, start low and increase until you hit resistance. I do this with everything - start at £500, if people buy easily move to £750, then £1,000, then £1,500. Find where people start pushing back, that's your ceiling.
Want the full unfiltered discussion? Join me tomorrow for the daily AI news live stream where we dig into the stories and you can ask questions directly.
Love AI with Kyle?
Make us a Preferred Source on Google and catch more of our coverage in your feeds.