These days almost every ecommerce tool claims to use AI: search, email, product descriptions, chat, analytics. If you run an online shop, that makes it hard to tell what actually changes something from what only changes the label.

A simple way to cut through it is to follow a purchase from start to finish, the way the customer experiences it, and ask at each step what genuinely gets better for them.

Before they reach your shop

The first change happens off your website. More visits now arrive from assistants such as ChatGPT, and it's starting to show: according to Adobe's data on US retail sites, visits arriving from AI assistants were up 138% year on year in May 2026, and converted 54% better than visits from other sources. The published figures don't say what share of total traffic they make up, so it's best read as a trend rather than the new main channel.

Here the AI isn't yours: the customer brings it. What does depend on you is what information those assistants find about your products, and that's decided in your catalogue. That's what getting your catalogue ready for AI covers.

Searching

Once inside, plenty of shoppers head for the search box. Semantic search makes a big difference to searches that describe a need ("comfortable trainers for standing all day") rather than a product. It's a real change, but a quiet one: the shopper still gets a list, just a better-ordered one. If someone searches for a specific model, ordinary search was already doing a good job.

Choosing

Inside your shop, this is where AI changes the most. Between seeing a list of products and deciding which one to buy there's a step ecommerce never fully solved: the one a good shop assistant handles in a physical store, listening to what you need, asking you something and pointing you to a couple of options. Filters, comparisons and 'you may also like' carousels help, but they don't ask questions.

An AI shopping assistant can: the shopper describes their situation in their own words and gets one or two recommendations for products the shop actually has, with a reason for each. It's the part where AI does something a website simply couldn't do before, which is why it changes the shopping experience most. It's also where you need the most care, because a recommendation that ignores the catalogue, stock or budget doesn't help, however convincing it sounds.

Checkout

At checkout, AI changes little, and that's fine. A good checkout is fast and predictable, and it doesn't need a conversation. Even OpenAI rowed back in March 2026: it admitted its first built-in checkout in ChatGPT wasn't flexible enough and now lets merchants use their own checkout, focusing on helping people discover products instead. What's happening with conversational commerce points more towards choosing better than paying differently.

After the purchase

After the sale, AI is mostly used in customer service: where's my order, how do I return it, when will it arrive. Done well, it saves hours of repetitive replies. For the customer, the difference is getting an answer straight away instead of waiting for an email, which is welcome, but it doesn't change how they shop. It's a different job from selling, as explained in chatbot or AI shopping assistant.

Behind the counter

Some of the most widespread uses of AI are ones the customer never sees: auto-generated product descriptions, campaign copy, sales summaries. It can save you a lot of time, with two caveats. A description written by AI without real data tends to sound like everyone else's, and if it invents a measurement or a feature, that's your problem. AI helps you write faster, but you still have to supply the facts about your products.

Where to start

If you can only pick one thing, start at the point where your customers get stuck. If you get lots of questions before people buy, or you see visits that open several product pages and leave without adding anything, the problem is choosing, and that's where AI can change the most. If what's swamping you is questions about orders, customer service is the priority. Either way, a well-described catalogue makes everything else work better.

SmartShop AI, a sales assistant for Shopify and WooCommerce stores, focuses on the choosing step: it works out what the shopper is after, asks a question if it needs to, and recommends one or two products from your catalogue, explaining why.