AI transparency
We advise our clients to declare where they use artificial intelligence, so we start with ourselves. This page says where we use it, what we send it and who makes the final decision. It is not a legal requirement: it is what we ask of the people who work with us. This is a translation provided for convenience: in case of any discrepancy, the [Italian version](/trasparenza-ia) is the legally binding one.
AI executes, the decision stays with a person
We use AI where it speeds up work a person has already decided on. We do not delegate the decision to it.
On the site
No model runs on certalea.com: no chatbot, no automated moderation, no automated decisions about the people who visit or write to us. The requests sent through the form are read by a person.
We write the texts of the site and of the articles ourselves, sometimes starting from a draft made with AI that we then rewrite. A person checks sources and figures before publication.
In our work
Our developers get help from Claude, from Codex and from models running on our own computers: they write and review code within work a person has already designed. How a piece of software is built and what it must do is decided by us, and the code that goes into production is reviewed by the person who signs it.
In this work, clients' personal data does not pass through the models: we remove it first, or that part of the work is done without AI.
In the AI systems we build for clients
An agentic system or an AI automation works on the client's data because the client chooses so. Before building it, we write down what data it reads, with which model and provider, what it can do on its own and when it hands over to a person. The client remains the data controller, and the processing is governed by the agreement pursuant to Art. 28 GDPR.
If this changes, we write it here
If a model were ever added to the site, this page and the privacy policy change before it reaches production, not after.