Shopify Says You Can Build a Custom Store in Twenty Minutes. Here’s the Asterisk.

TL;DR: Shopify launched Canvas on 1 October — an AI store builder where you describe what you want to Sidekick, its AI agent, and watch the shop rebuild itself. Its director of product says a merchant can now build a fully custom store in twenty minutes. That claim is from Shopify, and the same people say it’s “early stage” and “not replacing the existing editor yet”. The gap between those two sentences is the whole story.

What this AI store builder actually is

On 1 October 2026, Shopify launched Canvas — described by the company as “a design workspace that lets merchants build online stores with its AI assistant, Sidekick.”

The whole shop becomes one workspace. You see every page laid out together, pan across them, zoom in on a detail. Then you either click an element and edit it directly, or type what you want and Sidekick changes it.

The part that makes this an AI store builder rather than a smarter theme picker: it edits the real code. Not a mockup, not a preview — Canvas renders and produces actual theme code, with the interactivity and animation intact, which is why you can check how a page behaves rather than just how it looks.

What it replaces is worth understanding. Until now, changing a Shopify store meant rearranging modular sections and blocks inside a theme somebody else designed. Anything beyond that needed code, or a developer, or both.

The twenty-minute claim, and who is making it

This is the line that will travel, so here it is with its source attached.

Ben Sehl, Shopify’s director of product, founded the clothing brand Kotn. He frames it personally:

“When I built Kotn 12 years ago, I knew how to code and it still took me two weeks to get a rough store working.”

And now: “Today, an entrepreneur on Shopify can build a fully custom store in twenty minutes that’s head and shoulders above what I built then.”

Two weeks, by someone who could code, versus twenty minutes by someone who can’t. That is the case for an AI store builder in one sentence. If that holds up, it’s the clearest statement yet of what this category is actually selling — not code, but the fortnight you don’t spend.

But note who is making the claim. This is a company’s director of product describing his own launch, on launch day. It’s a credible person with relevant experience, and it is still marketing. Nobody independent has built a store in twenty minutes and reported back.

There’s one genuinely informative number underneath it: Sidekick made more than 25 million edits to themes in the first half of 2026. That isn’t a projection — it’s what already happened, and it’s why Shopify rebuilt its theme architecture so an agent could change store structure cleanly.

The asterisk, in full

Here’s the list every launch post buries and you need before you plan your week. As an AI store builder, Canvas does not yet support:

• Third-party themes — if you bought a theme, you may be out
• App blocks and app embeds — most of the apps bolted onto your shop
• Manually adding blocks
• Markets and market customisation
• Theme translations
• Theme updates
• Metaobject templates in navigation
• Market rollouts — a Canvas-edited theme can’t join one as an editable theme

It’s also desktop only, and it’s rolling out over the coming days rather than being switched on for everyone at once.

Shopify is refreshingly plain about this. Sehl calls Canvas “early stage”, says it “is not replacing the existing editor yet”, and concedes “there are gaps we still need to close.”

Read that list against your actual shop before you count on this AI store builder. If you run a translated store with three apps bolted on and a purchased theme, a twenty-minute rebuild is not available to you this week. If you’re starting fresh on a Shopify theme with no add-ons, it probably is.

Why this matters beyond Shopify

An AI store builder is a different proposition from an AI app builder, and the difference is money.

Most things our readers build with vibe coding tools are side projects, internal tools, or ideas being tested. A shop is a business on day one. It takes payments, it has customers, and when it breaks somebody doesn’t get their order. That raises the stakes on every one of these AI-generated changes.

It also lands in a week where we’ve seen apps built by non-developers drawing close to a billion visits a month. The category has stopped being about demos. This is the same shift arriving in commerce, where the feedback loop is immediate and measured in sales.

One thing worth carrying over from yesterday: we wrote about a critical flaw found in the framework underneath a lot of AI-built apps, and the lesson was that somebody has to watch the foundation you didn’t build. On a hosted platform like Shopify, that somebody is Shopify. It’s one of the real advantages of using a big platform’s AI store builder over assembling your own stack — and worth weighing against the flexibility you give up.

How to try it without breaking your shop

1. Duplicate your theme first. Work on the copy. Any AI store builder is safest pointed at something disposable, and this takes one click.
2. Check your blockers before you start. Third-party theme? App blocks? Translations? If yes, find out what Canvas will refuse before you’ve planned around it.
3. Start with one page. A single landing page or product page tells you how good the output is at a cost of twenty minutes, not a weekend.
4. Describe outcomes, not instructions. “Make this feel calmer and put the reviews above the fold” works better with an agent than a list of CSS changes. You’re briefing, not configuring.
5. Test the thing you can’t see. Click through to checkout on the copy. The rendered page looking right says nothing about whether buying still works.

Who should care (and who shouldn’t)

• Running a Shopify store on a standard theme with few apps: this is the most interesting week you’ve had in a while. Duplicate the theme and spend an hour.
• Running a complex store — apps, translations, multiple markets: read the unsupported list, then wait. Shopify says it’s closing the gaps; let them.
• Thinking of starting a shop: this genuinely lowers the bar. It does not lower the bar on having something worth selling, which is still the hard part.
• Building an app rather than a shop: nothing to do today, but watch the pattern. Every platform with a visual editor is about to get one of these, and our builder comparison is where that race gets tracked.
• On a different commerce platform: your vendor is building an AI store builder right now. The question to ask them is which of the eight unsupported things above their AI store builder will handle.

Our take

We think the twenty-minute number is probably true and almost certainly misleading — not because anyone is lying, but because of which twenty minutes it measures.

Twenty minutes to a store that looks custom, on a clean theme, with no apps: plausible. Twenty minutes to a store you’d actually put your name and your payment details behind — with the apps you depend on, the translations you need and a checkout you’ve tested — is a different number, and the unsupported list is how you can tell.

The Kotn comparison is the honest part of the pitch, though. Sehl could code, and it still took him two weeks. The value here was never that AI writes better code than a developer. It’s that the two weeks are gone, and for someone running a small business, two weeks is the whole difference between launching and not bothering.

What we’d do: duplicate the theme, rebuild one page, and judge the AI store builder on its output rather than on the claim. If a single product page comes back good, the twenty minutes is real for you. If it comes back generic, you’ve learned that cheaply — and the existing editor is still there, which Shopify is at pains to point out.

Not sure which builder fits what you’re actually making? Take the 60-second Vibe Coding Tool Finder quiz →

FAQ