跪拜 Guibai
← Back to the summary

Zero Hand-Written Code. One Hour of AI. A Fill-Light App Tops the App Store Paid Chart.

选择做什么、不做什么，永远是你自己的判断。

xiaomao-app-in.png

After all these pieces on tool usage and techniques, what you care most about is: how do you actually ship something?

This story shows how, without writing code yourself, you can produce a product that tops the paid chart.

(The following is in the author's first person)

The origin

One afternoon in November 2024, I was recording a Cursor tutorial video. My girlfriend sat next to me checking the shot when she suddenly said: "Instead of a screen flashlight, why not make a fill-light card?"

What is a "fill-light card"? I'd never heard the term. But I didn't ignore her; I opened Xiaohongshu and searched.

The results stopped me cold. Several posts about "fill-light cards" had hundreds of thousands of likes. Some people were using solid-color images as fill-light panels for selfies, others were comparing the effects of different colors, and some were looking for "a fill-light app that can adjust color and brightness," but none could find one that worked well.

I realized immediately: this is a genuine, high-frequency need. The demand was already validated; nobody had made the product.

Five minutes later, I opened Cursor (Claude Code wasn't being used yet) and typed a natural-language description of the features I wanted. An hour later, "Little Cat Fill Light" was live on the App Store.

The judgment

Before building, you first judge whether the need exists, then what to build. I did three things.

First, advice from a non-programmer.

My girlfriend doesn't understand code, but she understands user needs. "Fill-light card" is what real users call it, from Xiaohongshu. If I'd only talked with programmer friends, this direction would never have come up.

Second, three minutes of demand validation.

I didn't go on gut instinct that "this seems needed." I searched Xiaohongshu. Posts with hundreds of thousands of likes are the best market research report you can get.

Third, a minimal viable product.

The first version focused only on core functionality: a fullscreen solid color with adjustable color and brightness. No filters, timers, social sharing, or membership system. Just a fullscreen light you can tint.

The core rule: don't start by asking whether the technology can do it. First think about who will use it, in what scenario, and whether a good alternative already exists on the market. If many people are discussing a need and no good product satisfies it, it's worth building.

Building the iOS app

Create the SwiftUI project

First create an empty iOS project in Xcode (this step must be done manually, because Xcode's project initialization requires picking configuration in the GUI).

Once it's created, launch Claude Code in the project directory with this instruction:

This is a SwiftUI project, and I want to make a fill-light app. Core features:
1. Fullscreen solid-color display, users can pick the color
2. Adjustable brightness (from 0 to maximum)
3. A few preset color cards (warm white, cool white, warm yellow, pink, etc.)
4. Clean UI, usable the moment you open it
First look at the project structure, then implement the core features.

Iteration is the real work

The first version works, but it's far from "pleasant to use." User feedback tells you what to build next.

"Can you add a camera preview? Then I don't have to switch back and forth" → the core feature of the Pro version

"Color selection is too fiddly; can you make preset cards" → the color-card system

"I want custom colors" → HSB color picker

"Brightness is too low" → system brightness API + screen brightness overlay

Each iteration runs the same loop: user feedback → describe the need in Claude Code → implement → test → submit.

You might describe it like this:

Users say they need to see a camera preview while the fill light is on.
Add a feature: the top half of the screen shows the live camera feed,
the bottom half shows the fill-light color, with a draggable divider between them.

Claude Code will modify the existing code and add AVCaptureSession camera integration.

This is a technically complex feature (camera permissions, live preview, layout), but Claude Code handles it well, since these are standard SwiftUI and AVFoundation patterns.

App Store submission

Writing code was only 30 percent of the development process. The other 70 percent was submission work: registering a developer account, configuring certificates, writing a privacy policy, and waiting for app review.

Developer account

Releasing an app on the App Store requires joining the Apple Developer Program (around 688 yuan per year). Registration uses your real identity, and review usually takes one to two days.

Certificates and signing

This is the most counter-intuitive part of iOS development. You need to create certificates, configure provisioning profiles, and set up signing in Xcode.

You can have Claude Code lay out the full path from development to release:

I have an Apple developer account. Help me map out the complete flow from development to release, including certificate configuration, provisioning profiles, and Xcode packaging settings.
List the concrete steps for each stage and the pitfalls to watch out for.

Claude Code will give you a detailed checklist explaining which option to select and why, but it can't operate the interfaces for you. You'll do that manually in Xcode and on the Apple Developer website.

App review

App Store review is far stricter than the Chrome Web Store's. Common rejection reasons include:

"Little Cat Fill Light" passed review on the first submission, because its functionality was clear, its interface was clean, and there was no gray area. Review took about 24 hours.

Going viral: luck or necessity?

In the first week after release, I posted one note on Xiaohongshu. It was simple: showing the before/after contrast with fill lighting.

Within three days, it got 1.18 million reads and 73,000 likes. "Little Cat Fill Light" downloads passed 30,000. The "Pro version" (with the camera feature, priced at 6 yuan) topped the App Store's paid chart, holding the #1 spot for over a month.

Media later asked how I pulled it off. Honestly, even I was a bit surprised. But reviewing in hindsight, here are the key points.

The need is strong and unambiguous. Selfie fill lighting is a high-frequency, must-have use case. It's not "nice to have"; it's "can't do without."

The product is simple enough. Open the app and you get a fullscreen color card. No registration, no learning curve. Products that "work when you open them" spread most easily.

Controversy brings traffic. Many programmers commented things like: "This kind of app can make money?" "A thing built in an hour should be free?" The controversy generated even more attention and put the app in front of more people.

Fast iteration builds reputation. In the viral days, I pushed three versions overnight: fixing problems users reported and adding the most-requested features.

Practical expectation management. The vast majority of hand-rolled products die in silence. The success of "Little Cat Fill Light" contained an element of luck: the right track, catching Xiaohongshu's recommendation wave, and a product form that suits short-video distribution. None of those factors replicate every time.

The right product mindset: do every step well, but don't preset the outcome.

What actually matters

In this story, technology is the least important part.

Code, Claude Code can write. The submission process is documented online. Every platform has tutorials for promotion.

The factors that actually decide success are three non-technical ones.

Keeping your observation sharp in daily life matters more than any programming skill.

Focus on the product's core function. Don't bolt on unnecessary social features, memberships, or ads. Extreme simplicity is extreme competitiveness.

Going viral is a fluke; sustaining value is a choice. After "Little Cat Fill Light," I shipped several more versions, each solving real user-reported problems. That is the key to product success.

AI coding tools mean "writing code" is no longer the bottleneck. Competition shifts from "who can write the code" to "who can discover a problem worth solving."

Claude Code gives you the means to build anything you imagine. But choosing what to build and what to skip is always your own judgment.

For more quality articles as they're published, follow the WeChat official account 前端说书匠 (Frontend Storyteller).