Skip to content
Ernst Systems – Organize. Resonate. Navigate.

Legal

AI transparency

This English version is provided for convenience. The German version is legally binding.

Last updated: September 30, 2026 · Version 1

Orena is an AI system within the meaning of the European AI Regulation (Regulation (EU) 2024/1689, also known as the “EU AI Act”). Ernst Systems develops and offers Orena. Here we explain how Orena uses artificial intelligence and how we meet the transparency obligations.

Which AI powers Orena

  • Claude by Anthropic, currently a model from the Claude Sonnet family: understands requests, plans tasks, writes texts and, on instruction, analyses camera images and screenshots.
  • Gemini by Google, currently Gemini Flash Live: listens in live conversation and speaks with an AI voice. Gemini also helps with web searches, song recognition and in live camera mode.
  • Local models on the Mac: speech recognition (Whisper), wake word detection, speaker recognition and detecting when a sentence has ended. They work on the Mac without sending content to the internet.
  • The exact model version may change with updates. Claude and Gemini run on our customers' own accounts. We do not train our own AI models and do not use customer data for training.

Classification under the AI Act

Orena is a personal assistant for office and administrative tasks. In this use, Orena is not a high-risk AI system within the meaning of Article 6 and Annex III of the AI Act. The transparency obligations under Article 50 apply to Orena. Anyone using Orena in one of the areas listed in Annex III, such as selecting applicants or making decisions about employees, takes on their own obligations as a deployer.

How you know you are talking to AI

  • Orena introduces herself as an AI assistant. Every phone call starts with a notice such as “Hello, I’m Orena, the AI assistant of [name].”
  • Orena’s voice is an AI voice. It does not imitate any real person.
  • Orena drafts messages, emails and posts on behalf of her users. Users decide what Orena handles on her own and what they review first.

Labelling of AI-generated content

Orena does not remove markings that Anthropic and Google embed in generated content. Anyone who uses Orena to create images, audio or video that deceptively resemble real people, places or events, or who publishes texts without human review in order to inform the public, must label them as AI-generated under Article 50 (4) of the AI Act.

People stay in charge

  • After installation Orena starts in practice mode and sends nothing.
  • Before first use, users confirm in the app and during setup that they have taken note of the notices on the limits of AI. The confirmation is stored with the date and version.
  • Users decide which tasks Orena handles on her own and which ones she asks about first.
  • Orena only works with connected programs through the access the user sets up. Orena does not trigger payments.
  • Orena does not make solely automated decisions about people with legal or similarly significant effects.
  • Responsibility for approved actions remains with the users, as described in our Terms.

Limits of AI

AI results can be incomplete, outdated or wrong. Check important results before using them, especially for contracts, payments, legal or tax matters. Orena does not replace professional advice.

AI literacy

We continuously build our own AI literacy. During setup we explain how Orena works and where its limits are. Companies using Orena are themselves responsible under Article 4 of the AI Act for ensuring that their staff have sufficient AI literacy.

AI on this website

We wrote the texts on this website with the help of AI, reviewed them ourselves and take responsibility for them. If we show AI-generated images, we label them. The website itself contains no chatbot and no voice widget.

Questions

We are happy to answer questions about the use of AI in Orena: hello@ernst-systems.com.