Can AI copy
Shopify
Everything required to sell an object to a stranger, rented by the month.
Nice try
- Verdict
- Nice try
- Time to copy
- you won't
- Category
- ▮ Commerce
- Source
- Proprietary
The prompt
Written to be pasted straight into an agent with no editing. It describes the whole job, including the parts that will go wrong.
Do not build a Shopify. Build one store, for one merchant, on a headless commerce API you did not write, and keep tax to a dedicated tax service rather than a rates table you maintain. If you want to learn the domain, build the catalogue and cart: products with variant matrices and per-variant inventory, a cart that reserves stock transactionally at checkout rather than optimistically, discount rules with stacking precedence, and an order state machine covering partial fulfilment, cancellation and returns. That is the interesting part and it will take longer than you think.
What copies cleanly
- A storefront, a cart, a product catalogue with variants, and a checkout on top of a payment processor. This is a genuinely achievable fortnight.
- The admin: orders, inventory, discount codes, customer records.
- A theme system, if you keep the templating language simple.
Where it stops
- Sales tax. Rates vary by jurisdiction, by product category, and by the customer's location, they change constantly, and getting it wrong is a liability rather than a bug. This alone is a company.
- Checkout conversion is measured in fractions of a per cent and tuned over a decade. A checkout that is 3% worse is a merchant's entire margin.
- Inventory correctness under concurrency, across channels, with partial fulfilment, backorders and returns.
- Fraud screening, chargebacks, and shipping rate calculation against live carrier APIs.
- Black Friday. The whole platform's reputation rests on four hours a year, and there is no partial credit.
What actually protects them
- The app and theme ecosystem, which is a two-sided market that took fifteen years.
- Merchant switching cost: moving a live store is moving a business, mid-trade.
- Payments economics and fraud models that only work at their volume.
Editor's note
Commerce is where the gap between demo and production is widest. A store that takes money is an afternoon. A store you would let someone else's livelihood depend on is a different discipline entirely, and most of it is accounting.
Go and look at the original
We would rather you formed your own opinion than took ours. They built the thing; we only measured it.
Next in the index
- 1PasswordA vault for the secrets that unlock everything else you own.Nice try
- CraigslistClassified ads, in blue underlined text, largely unchanged since the last century.Nice try
- FigmaA design tool that runs in a browser and lets a room full of people edit the same file at once.Nice try
- GitHubGit, plus the social layer that made everyone agree on where the code lives.Nice try