How we use AI to build your online store

What AI does at Fullo

We use AI tools where they speed up work that can be checked. At the start of a project they help us turn your requirements, current site and catalog into an analysis and an architecture proposal faster: which pages, which product fields, which integrations and in what order. A person who has built similar stores then reads and corrects the proposal before it reaches you.

In development, AI writes part of the code: the groundwork for modules, connections to the bank, the courier and the ERP, and scripts that move data from your old store. A developer reads every one of those pieces, changes it where needed and approves it before it goes into the project. Nothing AI wrote reaches the site without that review. AI decides nothing on its own: it does what a developer asks, within the project, and every suggestion goes through the same review as code written by a person.

In content, AI helps product descriptions grow out of the attributes you already have, such as material, dimensions, color and purpose, and in search it helps the store understand synonyms and typos, so a customer who types “necklase” still finds necklaces. Before every release, AI also helps us write and extend automated tests for the cart, payment and order, so the same flows are checked on every release, not only before launch.

What stays with people

People review and approve every change. People also make the architecture decisions: which platform we build on, how the store connects to your systems, and what goes into the first phase and what comes later. The conversation with you and with the people who maintain your ERP is led by the team, not by a tool. The same goes for content: product descriptions suggested by AI are reviewed by someone on your team or ours before they are published, because a mistake in a description costs as much as a mistake in code.

AI speeds up writing, but it does not know your business: it does not know why certain customers get special prices, how your warehouse works or what happens when an order fails. That is why analysis starts with a conversation, not a prompt. Responsibility for the result is ours: if something does not work, the team that built the store and maintains it after launch fixes it.

What it means for you

For the same scope of work, the build time is shorter and the build price lower than when everything is written by hand. That is why a custom store is no longer the expensive option: on the Astro + Payload stack it fits the Start package, from €4,000, with a 5–8 week timeline.

The code is yours. Astro and Payload are open source, with no platform license and no lock-in to us: any team that knows these tools can take over and maintain the store, and we can if you want. The same level of checks applies to every release, because the tests we wrote stay in the project and still run when a new feature is added a year later.

Your data

Personal data of your store’s customers does not go to AI tools: names, addresses, orders and email addresses stay in the store’s database and in the systems it is connected to. Catalog data, such as product names, attributes and descriptions, goes to AI tools only when that is part of the job, for example when we write descriptions from attributes, and only when we have agreed on it.

An example: the new zlatara.net

We have been building the new version of the Zlatara Andrejević jewelry site on the same stack, Astro + Payload, since 6 October 2026. The site has 3,100 products and 16,700 images, card payments through NestPay, reservations, stock and price synchronization with the business system (AGV), feeds and two configurators, one for wedding bands and one for watch hands. Each of those two configurators used to be a year-long project.

The first version of the site took four days, the configurators ten, followed by three weeks of hardening and testing. That last part was not skipped to make the number look better: it is the full cost, which is why we include it in every estimate we give.

If you want to see what this would mean for your store, estimate your price or send us an inquiry.