Skip to content
valiss AI
EN DA

Responsible AI.

Compliance is not a fourth battleground. It is a layer in everything we recommend and build, from day one.

What we always do.

Customer data is stored in the EU
Hosting and backup in the EU. The data processing agreement comes with it from day one.
Data processing agreement before start
With us and with the technology partners that are part of the solution.
Least privilege
An AI solution never gets admin access. You own the key, everything is logged, and you can revoke it at any time.
AI says what it is
An assistant that talks or writes with your customers introduces itself as AI in the first sentence.
A person can always be reached
The handover happens at once when someone asks, with a short summary for whoever takes over.
Staff know what they use
Clear ground rules for what AI may be used for, and training, so the AI literacy requirement is covered.
Deletion by your rule
Conversations are stored as summaries, not audio, unless you choose otherwise and have a legal basis.
You own data and solutions
What we build, you can take with you. No lock in.

The rules, briefly.

This is an overview, not legal advice. Your specific setup should be reviewed with your data protection adviser.

  1. February 2025

    AI literacy for staff

    Companies must make sure that people who use AI know enough about it (Article 4 of the AI Act).

  2. August 2026

    AI must say so

    An AI that talks or writes with people must disclose that it is an AI (Article 50).

  3. December 2027

    High risk requirements

    Stricter rules for AI in recruitment and staff scheduling, among others.

  4. Always

    GDPR

    Data processing agreement, purpose, deletion and data in the EU, whether a person or a model processes the information.

Five principles we are measured by.

Read the full statement
  1. 01

    Accuracy over persuasion

    We report what is true, even when it is not convenient.

  2. 02

    Merit over manipulation

    We do not game AI systems, and we do not help others do it.

  3. 03

    Honesty about limitations

    AI makes mistakes. We say so instead of smoothing it over.

  4. 04

    Human accountability

    When our systems are wrong, the responsibility is ours, not the model’s.

  5. 05

    Responsible stewardship

    We collect only what the work requires, and use it for nothing else.

30 minutes. No preparation.

A list of your low-hanging fruit, and an honest view on whether you should start now.