AI-built ecommerce for schools: 40 stores live in three months | Kyron case study

Kyron sells technology and training to Italian schools. In May 2026 it was a brand new company: no brand, no site, no online sales channel. And a real deadline, June, because families buy their September materials over the summer.

That is where the analysis started. Three months later Kyron has a site, a visual identity, an operations studio and around forty dedicated school ecommerce sites, published by its own salespeople without a developer.

Starting point: a brand to build

Before any code, Kyron needed to be recognisable. We worked on the mark and the visual identity: how Kyron introduces itself to a head teacher, what tone it uses with families, what the material looks like in a teacher's hands.

Then the main site, kyronedu.it. Three non-negotiable goals: performance, SEO, and a clear experience for someone who lands from a Google search and has thirty seconds to decide whether Kyron solves their problem.

The site is not a brochure. Every page leads to an action: a shop, a request, a contact.

The studio: content run by agents

On top of the site we built Studio, Kyron's internal tool.

It had one simple job: make the team self-sufficient on content. No tickets to us to change a line of copy or publish a page. Inside Studio we put AI agents that assist whoever is writing, so a new catalogue or a new page ships without a developer.

That decision paid off later, when the hard part arrived.

The challenge: one ecommerce per school

Kyron works with many schools across Italy. Every school is its own case.

Specific products. Different bundles. Prices negotiated with that institute. Dedicated shipping methods. Sometimes a dedicated payment method too.

The traditional route would have been building one store at a time, by hand, with a developer per school. With a June deadline, that never stood up. And it would not have scaled afterwards either: every new school would have meant a new project.

Kyron needed a versatile system, not forty sites.

The solution: an AI agent that publishes the store

We built a single core, shared by every store, and put an AI agent trained inside Studio on top of it.

It works as a conversation. The agent talks to the Kyron salesperson and asks targeted questions — which school, which products, which bundles, which prices, which shipping, which payment — and generates interfaces on the fly so they can pick: product lists, discounts, options.

Then it publishes the store online. On its own.

Today a Kyron salesperson launches a school's dedicated ecommerce from scratch without calling an engineer. They do not edit a config file: they answer questions.

Results from June to August

The first store went live for the June deadline, in time for the summer buying season.

Between June and August Kyron published around forty dedicated stores. The digital channel generated hundreds of thousands of euros in revenue: not a launch spike, real sales to real schools.

The number matters. What matters more is that opening school number forty-one costs essentially nothing.

What we added after launch

Once the launch closed, Studio grew towards operations.

We added the analytics tools: orders, sales statistics per store, internal functions for the team's daily work. The people selling can see what is selling, school by school, without exporting spreadsheets.

What we take from this work

When a client has forty variants of the same problem, you do not build forty solutions: you build the system that generates them.

AI here is not a side feature. It is how a non-technical person does something that used to require a developer.

And the deadline is a design constraint, not a detail. June was not negotiable because families buy in summer. Everything else was shaped around that date.

The site is live at kyronedu.it.