Building a landing page with AI

A landing page is a single selling page for one product — the page an ad points at. You can build one yourself in the builder, or have AI build it and then edit whatever you want.

How the AI build works. It runs in stages: it reads your product's photos and words and analyses what sells it; you review and correct that analysis; it plans the sections; then it paints one picture per section. You are shown the plan before anything is drawn, so nothing is a surprise.

Two models, two jobs. One thinks and plans — it reads the product, finds the angle, writes every section and the brief for every picture. The other paints: one seamless picture per section, and on a capable model it renders the Arabic text inside the image itself. You pick each model, and you see the whole cost before you start.

Sections you tick. Hook, the problem, the solution, feature callouts, benefits, before/after, how it works, why choose us, reviews, guarantee, FAQ, countdown, and the price-and-button block. The order form, the summary and the address block are always added — it is a cash-on-delivery page, so it always has somewhere to order.

How many pictures. The planner decides, within the range the modal shows you: some sections become pictures, some stay text, and two sections can share one picture. The estimate says how many will be drawn and roughly how many credits that costs at most before you commit.

Colours. If you pick brand colours, the page and every picture are built on them. If you pick none, the colours come from your product's own photos (its colour, material and packaging), so a tan bag gets sand and olive tones, not a gold look on every page.

The buy button is part of the picture. The AI no longer adds a separate button block under the first picture. When the image model can write text, the opening picture and the closing picture paint their own buy button, and a tap anywhere on those pictures takes the shopper to the order form. With other models, a real button sits on top of the opening picture. You can turn the same behaviour on for any image in the builder with "Tap opens the order form".

While the pictures are drawn. A bar fills one picture at a time, and each picture is saved to the page the moment it lands. You don't have to wait on that window: close it with the × and the drawing carries on while you work anywhere else in the dashboard. A small progress pill sits at the top of the screen, under the top bar. Tap it to see the progress, stop the drawing (the pictures already drawn stay on the page), retry or skip a picture that failed, or open the editor. When everything is in, the pill says "Pictures ready" and stays until you open the editor or dismiss it. Only one page's pictures are drawn at a time, so a second AI page has to wait until the first one is done or stopped. Reloading the browser tab stops the drawing; what was already drawn is kept, and the editor can draw the rest.

It costs AI credits, and images are the expensive part — planning is cheap, painting is not. A page with no pictures is text-only and costs very little. Your balance and what a turn costs are in Settings → AI & models.

Nothing is locked in. The result opens in the builder where every section, word and image can be changed by hand, exactly like a page you built yourself. You can also re-run the analysis if the AI read the product wrong.

More in Online store

Still need help?

Ask the assistant from the search box in your dashboard. It answers from these same guides, in English, French or Arabic, and can make changes for you. Or write to us and a person will reply.