I'd argue the use case, but if it were legitimate this is actually the kind of thing where multimodal user interfaces seem promising to me. e.g. forms you can cooperate with a voice agent to fill out through a voice-first app