Last week at a café, Sister Lin asked me whether she should outsource her App
She's been on Taobao for 5 years and wants to move her customers into her own App. "Outsourcing quotes start at 300,000 RMB. I'm stuck." Neither of us had an answer that afternoon — but a Shopify news story keeps making me think this matters for us indie folks too.
So what is this Shopify news actually about
Shopify is one of the world's biggest e-commerce platforms — a lot of indie sellers use it to open their stores. 6 years ago they picked React Native — basically "write code once, get iOS and Android Apps both." But last month they suddenly announced: scrapping that decision, rewriting as native Apps, and the whole Shop App was done in just 12 weeks.
Why? Their official line is blunt: AI writes code too well now. Their engineering team found that AI writing native code (Swift for iOS, Kotlin for Android) is fast and stable, while React Native's "middle layer" became a burden. Note: Shopify isn't small — they employ hundreds of engineers. But AI let them ship the rewrite faster than they did 6 years ago.
What does this mean? "Building an App" is shifting from a big-company thing to something small teams can play with. I used to think making an App meant hiring outsourcers or learning to code for half a year. Now that thinking probably needs a refresh.
Tried it myself today — what's the actual cost
Money: For AI-writing-App yourself, the hottest tool is Cursor ($20/month, about 140 RMB). A domestic option is Trae, currently free.
Time: Don't expect it done in a week. But the most basic "display App" — putting your existing service into it — some people now finish in a few days. I personally got stuck on environment setup — installing all those dev tools on your computer — I messed that up twice before realizing Cursor's web version skips local install entirely.
Tech barrier: You need to tell AI clearly what you want. Honestly that's harder than coding, but also more valuable. You don't need to know what Swift is, but you do need to break down requirements.
First step: Open cursor.com, sign up, create a new project, and type into the dialog box: "I want to make an App. The homepage shows my product list, tap to see details." Let it run — even if it's ugly, that's fine.
Depending on your stage, here's how I'd think about it
If you're just starting out (no customers yet): Don't rush into an App. Get a mini-program or web version running first. App maintenance cost is high — wait until daily orders hit 100. It's fine not to try now.
If you have 1-2 steady customers: Use AI tools to make a prototype (something clickable, viewable) for the client to experience — not screenshots. I know a designer named Ah May who used Cursor to build a yoga studio booking App prototype for a client. The client signed on the spot — 10x more convincing than describing it verbally.
If you're scaling (team of 3-10): This deserves a serious talk. Pick the most curious person on the team, give them 5 hours/week to study AI coding tools. In 3 months you'll have gained a "half-technical co-founder." I once hired an ops person who couldn't code at all — with AI tools they now write their own automation scripts.