Our corporate policy for the use of artificial intelligence

Translations are for information purposes only. Only the German version, which you can find here, is legally binding: www.produktundprojekt.de/⁠ki.html

Fundamentals

Artificial intelligence (AI) is a technology that enables computers and machines to simulate human learning, understanding, problem-solving, decision-making, creativity, and autonomy. It therefore differs significantly from the unchangeable algorithms of traditional software applications. It is now no longer possible to draw a clear distinction between the use of artificial intelligence and manual processing using software applications. Artificial intelligence is firmly embedded in the operating systems of computers and communication devices, as well as in most online and offline applications, and often cannot be deactivated. We, too, use artificial intelligence within our company. Our staff have received specialised training in this area and have been made particularly aware of the potential for misuse of artificial intelligence. Our company adheres to the principle that artificial intelligence may only be used for the benefit of interested persons and customers. We comply with the regulations of the EU AI ACT, which have to be applied from 2 August 2026, and take into account the guidelines published on 20 July 2026 in this regard.

Use in the collection and verification of data

To ensure that quotations, order confirmations, invoices and shipping documents can be issued correctly and in compliance with legal requirements, we must verify the data provided to us – such as the company name, identification numbers, the billing address and, where applicable, the different delivery address – and complete any missing information. To this end, we use various online portals as well as internet search engines, such as the “Google” search engine operated by Google LLC, 1600 Amphitheatre Parkway, Mountain View, CA 94043, USA. Internet research also provides information generated by artificial intelligence. We are aware that this information may be incorrect. For this reason, this information is always cross-checked against at least one other source, preferably one that does not use artificial intelligence. Incorrect or misleading information generated by artificial intelligence is ignored and is neither processed nor stored.

Use in text generation

We do not use artificial intelligence to generate texts, such as replies to enquiries. We use artificial intelligence solely to check texts we have written ourselves for spelling and correct phrasing. Changes suggested by artificial intelligence – whether relating to spelling or phrasing – are not automatically applied, but must be manually approved by the author of the original text. We use artificial intelligence to translate texts into various languages, including some in which we do not have sufficient text comprehension. A second artificial intelligence system is used to check the spelling and review the phrasing of the translation. Furthermore, texts translated using artificial intelligence are always back-translated by the author into German, or, where appropriate, into another language with a very good understanding of the text. If a translation is ambiguous or contains errors, a new translation is produced until the result matches the manually created source text in terms of both content and style. Only after this is the translation further processed manually, for example by being pasted into an email. No artificial intelligence has automated access to our communication systems. When processing texts, we use the artificial intelligence “DeepL” from DeepL SE, Maarweg 165, 50825 Cologne, and the artificial intelligence “LanguageTool” from LanguageTooler GmbH, Boschstraße 23a, 22761 Hamburg. Both companies are based in Germany and are therefore subject to the General Data Protection Regulation (GDPR). This is important as texts, particularly emails, may also contain personal data. We do not produce any texts that are published for the purpose of informing the public about matters of public interest.

Use in verbal communication

We do not use artificial intelligence to communicate verbally. Telephone calls are always answered by a real person or by an answering machine that simply plays back a pre-recorded message. If communication in a foreign language is not sufficiently feasible, we will, where possible, ask the caller to send an email, to which we can then reply in writing.

Use in the editing of visual representations

Up to now, we have created visual representations based on photographs and by using conventional image editing techniques. For example, backgrounds were coloured completely white or flaws in the photos were retouched. Software that is not supported by artificial intelligence was used for this. The company owner is a graduate designer and took all the photos himself, editing them manually. However, artificial intelligence is now also being used in image editing. The aim of image editing is always to provide a realistic representation of the offered products, their features and potential uses. For example, modified product parts can be replaced in existing images. A photograph of the modified product part is used for this. This is therefore an AI-assisted image retouching process. The company owner and graduate designer, who is also the product developer, carries out this retouching himself, thereby ensuring that no images processed using artificial intelligence are used without his expert review. However, it is also possible to use artificial intelligence to mount existing images based on photographs into a fictional background, or to generate entirely fictional images. For image editing, the “Image Editor” from WorkSimpli Software, LLC, 1225 Ponce de Leon Ave., Ste. 1001, VIG Tower, San Juan, PR 00907, USA, in conjunction with image-editing software that does not utilise artificial intelligence. Attention is paid to ensuring that no personal data is contained in the images.

Labelling obligation

The guidelines on the EU AI ACT do not provide for a retrospective labelling obligation for content that was created or edited using artificial intelligence before 2 August 2026. Considering the explanations provided above, our company is not required to label texts or representations that have been manually edited using artificial intelligence after 2 August 2026. It is always a real person who creates, edits, checks and approves this content for use. Image retouching and changes to the background in promotional material are not subject to labelling requirements, provided that the material does not mislead regarding the product, its characteristics or its potential uses. Should there nevertheless be any content created after 2 August 2026 that requires labelling, it will be labelled accordingly. In the case of visual representations, this is done by a textual note, for example using the terms “AI”, “AI-generated”, “AI-edited” or the corresponding translations into other languages, or by using of the relevant symbols.