How it works

The main feature first, then the one for conversations you are already having. Nothing here is a mock-up — this is what the buttons actually do.

Real Life: the person in front of you

The main feature, and the reason the app exists. It works from a photograph of a situation you are standing in, not from a profile.

1
Photograph the scene, or describe it
Camera or gallery on a phone, gallery on a computer. If you would rather not take a photo at all, a written line about the situation works instead.
2
Faces are covered on your device
Before anything is sent. You see a light blur so you know who is marked; what leaves the phone has the face filled in opaquely. If no face is found the app asks you rather than refusing — you can say it is not visible, or tap it yourself.
3
Tap the person you mean
One tap, so there is no ambiguity when several people are in frame.
4
Enter age and mood as you read them
Two rows of chips, one tap each. You enter these. The app never works out age or emotion from a face — it receives them as your observation.
5
Get three opening lines and the context around them
A few seconds, 2 credits. Three lines in the style you picked, a sentence on when to speak and from where, seven topics to continue with, three questions phrased ready to say, and a reading of the scene. Your city and the time of day go into the request, so a Saturday evening and a Tuesday morning get different lines.

A worked example — a meetup on a Thursday evening:

That book has been on my list for about a year. Is it worth moving up?
Are you here for the talks, or did you just find the good chair first?
First one of these for me. Do people actually ask questions at the end, or is that a trap?

For conversations you are already having, in whichever app you are having them.

1
Screenshot the conversation
One to five images, in any order. They are sorted by the time they were taken, and one tap reverses that if you shot the conversation from the bottom up.
2
Drop them into Online Master
The text is read on your device and shown to you first, already with names replaced. You can edit it before anything is sent — the images themselves never leave.
3
Send one of the three replies
Written in the language of that conversation, in the style you picked, and shaped by what you have saved about the person. Copy the one that sounds like you.