AI transparency

Last updated: November 2026

Regulation (EU) 2024/1689 on artificial intelligence (the AI Act) governs how AI systems may be developed and used in the EU. This page explains what the regulation requires, where we use AI and how we implement these obligations.

This page supplements our terms (section 7) and our privacy policy (section 6).

1. What the AI Act requires

The AI Act classifies AI systems by their risk. The higher the risk to health, safety and fundamental rights, the stricter the obligations. It distinguishes providers, who develop an AI system (or have it developed) and place it on the market or put it into service under their own name, from deployers, who use it under their own authority. For our agent, we are both.

These rules matter for organisations like ours:

  • Prohibited practices (Art. 5, since 2 February 2025): for example manipulation, exploiting the vulnerability of people due to their age, rating people by their social behaviour (social scoring) and emotion recognition in the workplace or in educational institutions.
  • AI literacy (Art. 4, since 2 February 2025, amended by Regulation (EU) 2026/1744): providers and deployers support the AI literacy of the people who operate AI systems on their behalf.
  • General-purpose AI models (Art. 53 ff., since 2 August 2025): obligations of the providers of such models, in our case Mistral AI, for example on technical documentation and copyright.
  • Transparency obligations (Art. 50, since 2 August 2026): people must be told that they are dealing with an AI system unless this is obvious. Text, images, audio and video generated by AI must be marked in a machine-readable way. Deepfakes must be disclosed. So must AI-generated text meant to inform the public on matters of public interest, unless a person has reviewed it and someone holds editorial responsibility.
  • High-risk AI systems (Art. 6 and Annex III, from 2 December 2027 under Regulation (EU) 2026/1744): for example systems that decide on access to education, evaluate learning outcomes, select staff or allocate tasks based on personal traits or behaviour. Among other things, they need risk management, documentation, human oversight and registration in the EU database.

The General Data Protection Regulation (GDPR) continues to apply as well, in particular the right not to be subject to a decision based solely on automated processing (Art. 22 GDPR).

2. Where we use AI

We only use our own AI features in our agent. It uses language models from Mistral AI SAS (Paris, France) operated in the EU and has three tasks:

  • It classifies emails to agent@mail.stabilimkopf.com, for example as a workshop request, question, feedback, advertising or abusive message.
  • It suggests next steps to our team, for example asking trainers, proposing a date or sending a reply, and prepares drafts for them.
  • It helps our team look up details in the booking system in an internal chat.

The booking form, the login, the search for trainers nearby (we calculate the distance without AI), invoices and fee notes, the newsletter and this website work without AI. Our team holds editorial responsibility for the texts on this website and in our newsletter.

Stabily, the digital conflict coach, is a separate offer at stabily.ai and not part of this website. You can find the information about Stabily there.

3. How we classify our agent

Our agent does not use any prohibited practices. It is not a high-risk AI system either: it does not decide on access to education, does not evaluate students and does not select staff. For workshops, it only suggests whom our team could ask from a list of trainers drawn up by distance without AI; our team decides who is asked. Decisions about requests, dates and bookings are always made by a person on our team; its suggestions only prepare these decisions.

The transparency obligations (Art. 50) therefore apply to our agent; we also support the AI literacy of our team (Art. 4). We have documented this assessment and review it whenever the agent changes.

4. How we implement these obligations

For our agent we have put the following in place:

  • Notice of AI: the contact page says that agent@mail.stabilimkopf.com is handled by an AI-assisted system. Every email our agent writes itself ends with this notice; our other emails whose replies our agent reads say so in the footer.
  • Marking of AI text: replies whose text our agent prepared with AI are reviewed and sent by a person on our team. They carry a visible notice and, in addition, a machine-readable marking in the email header (X-AI-Assisted).
  • Fixed texts for automatic replies: confirmations of receipt, status updates and notices that our agent sends without a person are fixed texts written by us. The AI only decides whether they are sent, and only to technically verified sender addresses. In addition, the agent automatically blocks sender addresses for agent@mail.stabilimkopf.com when the AI classifies their messages as abusive, repeatedly as advertising or repeatedly as off-topic; our team can lift blocks at any time.
  • Human oversight: only our team carries out suggestions, after changing them if needed. Automatic replies can be switched off one by one, and the agent can be paused at any time. Every action is logged.
  • AI literacy: everyone on our team who works with the agent gets internal guidance on what the agent can do, where it makes mistakes and how to check its suggestions.
  • Data protection: Mistral processes the data as our processor in the EU and does not use it for training. Details are in our privacy policy (section 6).

5. Your options

If you do not want our agent to process your message, write directly to our team: wirsind@stabilimkopf.com. Our team handles emails to this address itself; our agent does not evaluate them. Please note: replies to our emails about your request go to our agent.

We are happy to answer questions about our use of AI at the same address. You can also lodge a complaint with the competent market surveillance authority (Art. 85 AI Act); your rights under the GDPR are described in our privacy policy.

6. Changes

We update this page when we use AI differently or when the legal situation changes, for example through guidelines from the European Commission.