Everyone has a conversation they are dreading this week.
The call to book a doctor’s appointment. The $35 fee you let the bank keep. Asking for $18 an hour. The extension you never asked for. The group project partner who went quiet. The $72 a friend still owes you. Telling your mom you’re not coming home. Saying no to the trip.
Sixty-five percent of Gen Z say calling a stranger makes them uncomfortable. Only a third are comfortable making calls at all.1 This is not a character flaw. It is a skill nobody was given a place to practice.
Language apps figured this out years ago: get people to say it out loud, every day, and reward the rep. Speak built a hundred-million-dollar business on that.3 Nobody built it for the conversations people actually fear, in their own language.
We are not a companion.
AI companions are built to keep you talking to them. Their success metric is your minutes. Ours is the opposite: the conversation you go have after you close the app.
So there is no romantic roleplay. No endless chat. No “I missed you.” Every persona is clearly synthetic, every rehearsal has an end, and every end has a debrief that points you back at a real person.
A coach, not therapy.
Unmute is practice, not therapy or medical care. If crisis language shows up in a rehearsal, we stop rehearsing and route to real resources, every time.
Your rehearsals are yours.
Rehearsals are private to you, never used to train models, and deleted when you say so. Unmute never listens to real calls. Phones don’t allow it, and you wouldn’t want it to. Sharing a clip is always your explicit choice, voice-changed if you like.
Three minutes, every day, until you don’t need us.
The whole product is a rep and a debrief. The other person pushes back. You make the ask and stop talking. You get told exactly what to change and the two sentences to try next time. Then you go say it to a human.
If it works, you will use Unmute less over time. That is the point.
The Unmute team, September 2026