For our client from Southern Germany we designed, extensively tested and now deployed to the cloud a production system which each day transforms ten-thousands texts within a few hours and reports them back. The client sent short texts are automatically transformed using ChatGPT from OpenAI. As part of this process ChatGPT answers are linguistically checked for undesired words whose presence triggers regeneration of such texts. To reach the required processing speed while staying below load limitations, we use a high degree of parallelization on different architectural levels.