MAUTZI. Technology that listens.

In residential care there is little time for conversation and activity. Care staff and residents described this consistently in our conversations on site.

Digital offerings often fail with exactly this audience because they assume screens, menus and technical knowledge.

I founded MAUTZI together with Phine, supported by the Lower Saxony Startup Grant, administered by NBank from January to October 2026.

My focus is product and user research. I run the conversations in the care home, translate the findings into product decisions and model the service processes in my bachelor's thesis.

We first spoke directly with residents and care staff in a care home. The insight was clear. Interest was not the main barrier, ease of use was.

That led to the central product decision. MAUTZI runs as a tablet app in kiosk mode, an animated cat embodies the assistant, and it is operated by speech. One button starts the conversation, then voice activity detection takes over, no wake word needed.

Technically MAUTZI is a client-server system with a server-side persona that keeps answers to one to three sentences, German only. Free conversation runs in realtime over OpenAI, the voice games use a grammar-constrained recogniser with Vosk.

We then tested the prototype on site with residents instead of judging it at a desk.

There is a working prototype that was tested on site with residents.

It covers free conversation, appointments, weather, news, a media library, voice games like bingo and memory, and a drink reminder every two hours.

The test produced the requirements for the next stage, and my bachelor's thesis contributes the modelled service processes alongside it.

The prototype works and was tried out on site with residents in a care home.

MAUTZI is not yet in regular operation. Without measurement there is no evidence of any effect on wellbeing or loneliness, so we do not claim one.

MAUTZI complements human care rather than replacing it. We know that voice control is not accessible to everyone. People who cannot or do not want to communicate by speech need other routes, which is why alternative input and fallback layers are the next step.

Closeness to the people you build for beats any assumption made at a desk. Almost every good decision came from a conversation on site.

Today, conversation, games and queries run by voice, while starting still takes a button press. Next up are voice operation without any button, a reliable fallback and a longer trial in daily operation.