Do you need to re-platform to get AI search?

You've seen the AI search demos. Someone types "something for a beach wedding" and the right dress appears. Your search returns nothing for "beach wedding" because none of your product titles contain those words.

So you ask the obvious question: do we have to move off our platform to get that?

Almost always: no. And you should be suspicious of anyone who says yes without asking what else is wrong.

If AI search is the only problem, buy an app

If you're on Shopify or BigCommerce and your only complaint is search, there are apps for this. Searchspring, Klevu, Rebuy and others have been doing semantic search and recommendations for years. They install in an afternoon, they cost a few hundred a month, and for a catalog under a few thousand SKUs they will get you most of the way.

Re-platforming to fix search is like moving house because the shower pressure is bad. Technically it works. It is not the cheapest way to get a better shower.

We build AI-native commerce platforms, so this is us telling you not to hire us for that. If search is the whole problem, an app is the right answer and we'd rather say so now than eight weeks into a project you didn't need.

The question underneath the question

Here's what usually happens. Someone comes to us about search, and forty minutes in it turns out search was the thing they could name. The actual list was longer:

• Checkout can't do the thing their business model requires — subscriptions, bundles, deposits, split shipments
• B2B pricing is being faked with customer tags and a spreadsheet
• The storefront is slow and every fix is fighting a theme they don't control
• They're paying for eleven apps that each own a piece of the same customer record
• They want to launch in another market and the platform's answer is "a second store"

Any one of those is worth a conversation about the platform. Search isn't. But search is the visible symptom, so it's the one that gets raised.

The honest test: if AI search landed perfectly tomorrow, would you be happy with your stack? If yes, buy the app. If you'd still be annoyed about four other things, the platform is the constraint and search is just where you noticed it.

Where our AI actually runs — and why we say so

Our search, recommendations and shopping assistant are modules on our own commerce engine. They read the catalog, cart and order data through it. There is no version of them today that installs into a Shopify store.

We used to describe this as dropping into your existing store. That was wrong, and we changed the copy. It's the kind of thing that's easy to write, sounds good in a pitch, and falls apart the day someone says "great, install it".

So when we build a platform, the AI is already in it — that's why the AI phase takes a fortnight instead of months; we're fitting something that exists to your catalog rather than writing it. But you're getting the AI because you're moving, not moving in order to get the AI. That order matters, and reversing it is how people end up with a rebuild they didn't need.

What to check before you scope anything

Whether you move or install an app, one thing decides whether AI search works for you, and it isn't the vendor.

Print the text you're indexing — title, description, attributes — for five products, side by side. If they read like the same marketing sentence with the nouns swapped, no search product on the market will save you. We watched our own vector search rank sweatpants first for "breathable for a hot summer day", because every description in the catalog was the same paragraph with one word changed. The fix was the copy, not the engine. We wrote that one up separately, numbers and all.

That check costs you ten minutes and it's the same answer regardless of who you buy from. Do it first.

The short version

• Only search is broken? Buy an app. Don't re-platform.
• Search plus four other things? The platform is the problem. Scope that, and the AI comes with it.
• Either way, look at your product copy before you sign anything.

If you're in the second bucket and want a second opinion on whether you actually are, that's what our architecture workshop is for — it's free, and about a third of the time the answer is "stay where you are and fix these two things".