I record calls, transcribe them locally, then paste the transcript into a model with a fixed prompt asking for decisions, owners, and deadlines. It works, but the output still needs heavy cleanup and I re-explain context every single time. Curious how others handle this. Do you keep a running project doc the model reads first, use a strict template, or just accept the manual fixes?
What's your actual workflow for meeting notes to tasks?
Replies
No comments yet — be the first to share your thoughts.