一個非工程師,在 App Store 上架了 30 幾隻 app
A Non-Engineer Who Shipped 30+ Apps
🇹🇼 中文
一句話
我沒寫過一行「會上架」的程式。但現在 App Store 上、有三十幾隻 app、是我一個人、從想法、寫程式、過審查、到上架、自己走完的。
問題
做工務這些年、我腦袋裡一直有一堆「要是有個小工具就好了」的念頭:磁磚要鋪幾片、地板損耗怎麼抓、繩結怎麼打、工地日誌怎麼記……
但對一個非工程師來說、「自己做一隻 app」聽起來像天方夜譚——學 Swift 要多久?上架那套規則誰看得懂?光是想就放棄了。
直到我發現、AI 可以把這條「不可能」的路、變成一條「一步一步走得完」的路。
怎麼用 AI 落地
我把「做一隻 app」拆成一條我能跟著走的流水線、每一段都讓 AI 當我的副手、但每一個決定都還是我下:
- 從現場痛點長出 idea → 想清楚這隻 app 到底解決誰的什麼問題(這步最關鍵、也最不能交給 AI)。
- AI 協作把功能寫成 Swift → 我看得懂邏輯、改得動細節、卡住就一起 debug。
- 走完 Apple 的整套上架 → 審查、隱私頁、內購、截圖、送審——這些「魔鬼細節」我親自踩過、也踩過坑(版本遷移、審查被退、加密宣告…)、一隻一隻磨出經驗。
- 變成一條可複製的產線 → 第一隻最痛、到後面、我有了自己的上架 SOP、出一隻的成本越來越低。
量化結果(可驗證)
- 32 個 app 專案、其中 7 隻以上已在 App Store 上架(可查連結)。
- 橫跨領域:裝修工具(鋪磚 / 地板 / 丈量)、工地工具(工地日誌 / 進度)、生活工具(植物辨識 / 高爾夫 / 結繩)。
- 不是「一隻試水溫」、是一整批——從 0 到 App Store 的完整流程、我跑了三十幾遍。
這個案例證明什麼
非工程師、也能獨立交付通過 Apple 審查的真實產品、而且能做到「量產」級。這證明的不是「我會寫程式」、是「我能用 AI 把一件原本要靠專業團隊的事、變成一個人扛得起的流程」——這正是 AI 落地最核心的能力。
🇬🇧 English
One-liner
I had never written a single line of "shippable" code. Yet today there are thirty-something apps on the App Store that I took from idea to code to review to release — alone.
The Problem
Over my years in site work, my head was always full of "if only there were a little tool for this": how many tiles to lay, how to estimate flooring waste, how to tie a knot, how to log a job site…
But for a non-engineer, "build your own app" sounds impossible — how long to learn Swift? Who can even read Apple's release rules? Most people quit at the thought.
Until I realized AI could turn that "impossible" path into one you can actually walk, one step at a time.
How I Landed AI on It
I broke "build an app" into a line I could follow, with AI as my co-pilot at every stage — but every decision still mine:
- Ideas grow from real on-site pain → get clear on whose problem this app actually solves (the most important step, and the one you can't hand to AI).
- Co-write the features into Swift with AI → I can read the logic and change the details; when stuck, we debug together.
- Walk Apple's full release process → review, privacy pages, in-app purchase, screenshots, submission — the devil's-detail steps I've done by hand, and hit the potholes on (data migration, rejections, encryption declarations…), grinding out experience app by app.
- Turn it into a repeatable pipeline → the first one hurts the most; by the end I had my own release SOP, and the cost of shipping one kept dropping.
Measurable Results (verifiable)
- 32 app projects, 7+ already live on the App Store (links available).
- Across domains: renovation tools (tiling / flooring / measuring), job-site tools (site diary / progress), lifestyle tools (plant ID / golf / knots).
- Not "one to test the water" — a whole batch. I ran the full 0-to-App-Store process thirty-something times.
What This Case Proves
A non-engineer can independently ship real products that pass Apple's review — at production scale. This doesn't prove "I can code"; it proves "I can use AI to turn something that used to require a professional team into a process one person can carry" — which is the very core of landing AI.