What happens to your brief.

You do. Everything an agent produces on your project belongs to you once the work is paid for, including the drafts you did not use. We keep no licence to reuse it, and we do not show it to anyone without your written permission.

What we cannot promise is that a generated image or line is free of every third-party claim. We tell you which outputs came from which model, so your legal team can apply its own rules.

Not by us. Aixent never uses your briefs, brand guidelines, campaign data or outputs to train or fine-tune a model.

The model providers we work with set their own terms for what they do with inputs. They are named below, so you can check each one against your own rules.

Two lists, because they are two different things.

On this website

The newsletter is sent through Amazon SES. The privacy notice has the full list.

Where your project files live

On your own servers. We install the agents on your infrastructure and hand them over, so your briefs, guidelines and outputs are stored with you, not with us.

What we hold ourselves, during the build and in Lab sessions, sits on our own server at Hostinger in Germany.

When an agent works on your project

Wherever the agents run, they send prompts to one or more of the model providers below.

We tell you before a new provider touches your project. If there is a provider or a country you do not want, tell us and we leave it out of your project.

Each client has its own workspace. Nothing from one client's project is used as context for another's. We delete project material 90 days after the engagement ends, or on the day you ask.

People do. Agents draft, compare and recommend. A person signs off at every handoff, because taste is still the job. No agent pretends to be a person.

Nothing goes to a client or a channel without that sign-off, and the name of the person who gave it stays with the work.

It will, sometimes. Every language model does.

So research comes back with its sources attached, and a claim without one is flagged before handoff. Numbers in a report are calculated from your data, not written by a model. The governance step checks brand, legal and factual points before anything moves on.

You keep every output, along with the prompts, configuration and brand material we built for you, in formats you can open without us. We delete our copies on the schedule above.

A person from the studio runs every session. What you bring is used for that session only and deleted when it ends.

We hold no security certification: no ISO 27001, no SOC 2. We are a small studio and would rather say so than imply otherwise.

We sign NDAs and data processing agreements, and we answer security questionnaires.