Feature request for product/service
Chat
Describe the request
One UX issue I keep running into is with voice input.
Often I’ll spend several minutes dictating a detailed prompt while another agent is working in the background. When that other task finishes, I instinctively click over to it before I’ve hit Send on my voice prompt. The problem is that the draft is then discarded, meaning I have to dictate the entire thing again.
Voice is by far the fastest and richest way for me to communicate with Cursor, so losing a five-minute dictation is incredibly frustrating.
I’d love to see voice input behave more like a persistent draft:
- If I’ve started a voice recording, don’t discard it just because I switch to another task or agent.
- Keep the draft until I explicitly send it or delete it.
- If processing is still happening, preserve the conversation state so I can come back and continue where I left off.
The ideal experience would be that voice input is treated as something valuable that shouldn’t disappear accidentally. Switching context between agents should be seamless without the risk of losing several minutes of dictated work.
Does anyone else run into this, or is it just the way I naturally multitask?