Get started

An AI shop assistant that sells — with today’s prices and real stock

Filters and search bars can’t ask “what is it for?”. A good AI assistant can — and recommends from your live catalogue, with clickable product cards, without ever inventing a price or an “in stock”.

A Scandinavian-style living room with a floor lamp and armchair, a smartphone on a side table showing a shopping chat

In a physical shop, the best salesperson doesn’t start with the catalogue. They ask: what is it for? Where will it go? What’s your budget? Then they walk you to two products, not two hundred.

Most webshops can’t do that. They have filters, a search bar and a contact form — and a visitor who doesn’t know the right keyword leaves. An AI shop assistant brings the salesperson back: it asks the questions, understands the answer in plain language, and recommends from your actual catalogue.

What shoppers really ask

The questions in a webshop chat fall into two groups. The first is product advice:

  • “I need a floor lamp for a reading corner, under 40,000 Ft.”

  • “What goes with an oak dining table?”

  • “Is this one available in grey? Is it in stock?”

  • “What’s the difference between these two?”

The second is everything around the purchase: shipping time and cost, returns, warranty, payment methods, invoicing, pickup. Both groups have one thing in common: the answer is worth money only if it is correct today.

The Nordik Otthon demo shop’s assistant recommends the Fjord floor lamp with a product card showing 34,990 Ft and in stock, then says the Strand rug is out of stock
A recommendation from the live catalogue, with a clickable product card — and an honest “out of stock”.

Product cards, not paragraphs

When the assistant recommends something, it doesn’t just describe it. It shows a product card — photo, name, price, stock status — that links straight to the product page. One click, and the visitor is where the “Add to cart” button is.

It also keeps the constraints the shopper gave. If they said “under 40,000 Ft” and “for a small flat”, the second and third suggestions respect that too, instead of drifting towards the most expensive item in the category.

How does it know today’s price?

The catalogue doesn’t live in the model’s memory. It is synced straight from your shop: there are ready-made connections for WooCommerce, Shopify and UNAS, and any other platform works through a product feed or our product API. Prices, stock, sizes and materials update by themselves — no copy-pasting, no forgotten price change.

For every question the assistant searches the catalogue with real filters — budget, colour, size, category — and only sees the matches. Even a shop with thousands of products doesn’t drown it in its own range.

Diagram: your shop (WooCommerce, Shopify, UNAS, product feed) → automatic sync → live lookup with real filters → product card; if something is not in the data, the assistant says so
The catalogue comes from your shop and is looked up fresh for every question.

The rule that matters most: no invented price, no invented stock

Here is the uncomfortable truth about cheap shop chatbots: a language model fed with page text will, sooner or later, produce a price that isn’t yours or write “in stock” about a product nobody has checked. Not out of malice — it is simply what language models do when they fill a gap.

Enéh treats this as a hard rule, not a style preference. Prices and stock come from structured product data, looked up for each question. Before an answer is sent, any amount or stock claim that doesn’t appear in your data is removed. If the stock status is unknown, the assistant says so and offers to check with the team.

Out of stock is a sales moment, too

“Sorry, it’s out of stock” is where most shops lose the customer. The assistant says it plainly — and in the same breath offers the closest alternative that is available, or a way to be notified. A visitor who came for a sold-out rug can still leave with a rug.

Beyond products: the questions that block the checkout

Many abandoned carts aren’t about the product at all. They are about a doubt: Will it arrive before the weekend? Can I return it if the colour is wrong? Can I pay on delivery? The assistant answers these from your own shipping, returns and payment terms — the moment the doubt appears, not two days later by email.

And when a shopper needs more than a chat — a styling consultation, a showroom visit, a custom order — it can book that appointment or pass a qualified request to your team.

What to measure

  1. Clicks from product cards to product pages.

  2. The questions asked most before a purchase — they show what your product pages are missing.

  3. Out-of-stock questions: which products people want that you don’t have right now.

  4. Conversations handed to your team, and what they were about.

See it in a real shop

Our demo store, Nordik Otthon, sells Scandinavian furniture and lighting with a live catalogue — some items in stock, some orderable, one sold out. Ask it for a lamp under 30,000 Ft, then ask about the rug that isn’t available. Open the webshop demo. Running WooCommerce, Shopify or UNAS? Request a demo on your own shop.

Magyarul olvasnád? AI asszisztens webshopba: termékajánlás valódi árral és valódi készlettel.

See it working

Seven industry demo sites, each with a trained Enéh assistant on it — ask them anything.

Browse the demos