AI Impact · Principles

Unic AI Principles and Guidelines. How we use AI, and where we set boundaries.

We use AI to create real added value for employees and customers without compromising security or data protection. This page reproduces the principles and guidelines in full.

Process P12.5 Last reviewed 20.07.2026 Owner Ivo Bättig
Source of truth. This text is maintained as the TQMI process P12.5 — Unic AI Principles and Guidelines. It is reproduced here verbatim and is not edited on this page; in case of doubt the TQMI process prevails.
The process, quoted

Purpose

In the world of professional services, artificial intelligence (AI) is fundamentally changing the way we work. As a relatively young foundational technology, AI is still in an experimental phase in many areas. That is why we explore and test both the opportunities and the limits of this technology. At Unic, we use AI internally to support our work and in our customer projects.

We use AI to create real added value for employees and customers without compromising security or data protection [1]. These principles and guidelines describe how we use AI meaningfully — and where we deliberately set boundaries.


The use of AI should increase efficiency and promote innovation. To comply with security standards, ethics and legal requirements, we pay attention to confidentiality, transparency, responsibility and sustainability. The use of AI at Unic is based on the following four principles:

Confidentiality

Protecting customer data and sensitive data is a priority. We process sensitive information only in approved AI tools.

Responsibility

Anyone who approves AI-generated content assumes the same responsibility as for content they created themselves — towards the customer, towards Unic and towards third parties. We therefore check generated results for accuracy, usability, ethical guidelines, fairness and possible bias.

Transparency

We disclose when, where and how AI is used in projects, both internally to colleagues and externally to customers. We always label content or processes that were mainly generated by AI as such and, where appropriate, document in a traceable way which prompts, model versions and data sources we used.

Sustainability

We take responsibility for the ecological impact of using AI. That is why we prioritize lean models, efficient prompts and resource-saving architectures to reduce energy consumption. We integrate this awareness into every project and every use case to advance Unic's commitment to sustainable digital solutions.

Unic AI Guidelines

We actively integrate AI into the way we make decisions and solve problems. We encourage experimentation with AI to understand its capabilities and limits and to develop innovative applications. However, we must ensure that AI is used responsibly. To make this possible and apply our AI principles in everyday work, we follow these four simple checks:

1 Data check

The following requirements apply to the processing of internal and customer data: Approved tools and data classes (an integral part of these guidelines).

2 Content check

We always review AI results ourselves (for larger deliverables, also by a colleague) and check facts, tone, brand language, bias, copyrights and trademark rights. A human must review and approve every output.

3 Labeling check

We want to communicate as openly as possible about the use of AI technologies. Therefore, label AI content internally and externally. In Confluence/Teams, if content was mainly created by AI, label it as "AI-generated" (AI created the content, a human reviewed and approved it) or "AI-assisted" (a human created the content, AI assisted with research, corrections or ideas). For external deliverables, label it as "created with the support of AI".

4 Sustainability check

Use AI where it makes sense. Avoid unnecessary iterations. Use efficient architectures and small models for projects. We measure AI use internally and in projects and integrate it into our carbon footprint.

[1] General Data Protection Regulation (GDPR) – European Union, Federal Data Protection Act (BDSG) – Germany, AI Act (future EU AI law), ePrivacy Directive (EU), revDSG – Federal Act on Data Protection (Switzerland), RODO (Rozporządzenie o Ochronie Danych Osobowych), CCPA (California), PIPEDA (Canada), LGPD (Brazil) — if data is processed internationally — and copyright law.

Knowledge

Questions? Ask the AI Impact agent

AI Impact Agent

The Knowledge section is not duplicated here. Ask the agent — it knows the whole programme and forwards what it cannot answer. Responsibility remains with the employee.

Ask the agent →