The more I use AI agents, the less I want to type long prompts on a laptop.
Spoken, messy English is usually fine — the model gets it. On a phone, voice input is easy. On a desktop, it still hurts when the mic is bad, far away, or missing. Built-in OS dictation helps if you have a decent mic. When you don't, people often just go back to the keyboard.
Another path I've been using: talk on the phone, and the text lands in whatever box has focus on the PC — chat, Cursor, the browser, etc. That fits agent workflows well because models already handle colloquial speech; you don't have to polish it into formal writing first.
That's what FlowMic is for: speak on the phone, words show up at the PC cursor.
Curious how others actually talk to agents day to day — keyboard only, desktop dictation, or phone-to-cursor?
