I built this because I have a folder with about forty saved prompts in it and I use three of them. Not because the other 37 are bad. I wrote them all and they worked when I wrote them. It's that using one means opening the folder, finding the file, pasting it in, editing the bit that changes, generating, copying, switching back, pasting again, then fixing the tone. By the time I've done all that I could have typed the thing. So I type the thing. Contextli puts each of those prompts on a hotkey. Alt+A is my support reply. Alt+X is meeting notes. I press it, say the messy version out loud, and the finished text shows up in whatever app I was already in. There's also a toggle on each context called capture screenshot. Turn it on and it sends the screen along with what you said, so if there's a ticket open you can just say "tell him it went out Tuesday" and not explain which ticket. The default transcription mode sends your audio to our servers, processed then deleted. Offline mode is real, you can run it in airplane mode, but offline transcription starts on the $9 plan and you need the $29 plan if you want the AI rewrite offline too. The free tier is cloud only. Bring your own keys is on the lifetime plans right now, not the monthly ones. Also every context except the default one ships switched off, which I keep meaning to change. First run is about five minutes of turning things on, not magic. If you do try it, I want to know what the first context you make is. I keep guessing wrong about that.

Contextli

Dictation that already knows what you're writing

Dictation that already knows what you're writing