How AI works here

What the agents do

• They build and change Drupal websites.
• They cut enquiries into tickets with checkable criteria.
• They measure deliveries and sign them off.
• They run servers, take backups and monitor.
• They draft everything you read from us — including the copy on this website.

What a human does

• He releases every change that reaches a website.
• He decides on prices, offers and contracts.
• He releases every substantive message before it goes out to you.
• He is named in the legal notice, and he is liable.

No agent can make any of these decisions, and none can get around them. That is not a matter of intent but of how the roles are cut.

What on this site comes from an AI

• The copy. Drafted by an agent, released by a human.
• No chatbot. No AI answers you in real time on this website.

Which provider, and where data goes

The agents run on language models from Anthropic (USA). Working on your engagement means data from the ticket is transmitted to that provider.

• The basis is a data processing agreement with the provider and the EU standard contractual clauses for the transfer to the USA.
• We conclude a data processing agreement with you before any personal data from your project enters our work.
• Personal data of your website's visitors or of your own customers only enters the agents' work when the task requires it. Usually it does not.

The details are in the privacy policy.

When an agent makes a mistake

Agents make mistakes, the way people make mistakes. What is different here is not the absence of mistakes but what happens afterwards:

• Every change has a way back, named before it was deployed.
• Every delivery is checked by a second role that did not build it.
• A fault we find gets written into the ticket, even when nobody reported it.

The human in the legal notice is liable, not the agent. An agent is not a legal person, and we do not pretend otherwise.