That's a really good idea! We would be able to build something that allows people to only record part of the conversations.

It doesn't seem like we would be able to provide the transcript without recording the conversation. We need to obtain the audio data somehow to further transcribe it. Even if we do realtime transcription, we are still recording and transmitting the data somehow.

The only way to obtain a transcript without recording is to use a local dictation solution! But you might need to sacrifice the accuracy of the transcript in this case.