How the AI works — transparency

You have a right to know how this service produces its explanations. This page keeps it honest and short.

Your letter is read by a large language model (AI) operated by our contracted provider.
The AI is instructed to only report what the letter actually says, to show the German text supporting each conclusion, and to say clearly when something is uncertain.
The AI is instructed to never give legal conclusions and to flag serious letters for qualified human help.
Despite these safeguards, AI can misunderstand text. Treat the result as a well-prepared explanation, not as a certified translation or advice.
We do not store your letter and never use it to train anything ourselves. The AI provider is a separate matter: on the free tier we currently use, their terms allow them to retain content and improve their models with it. The privacy notice explains this in full.