Four people, and the room for many more.
Each of them carries their own name, their own written character, their own pronouns, their own hellos and their own default voice — in their own file, not in the app. Swapping the model swaps the person. Click a face to read who they are.
A mood is layered on, not swapped in
Twelve directions — Infatuated, Loving Daughter, Tsundere, Little Gremlin, Insufferable Genius, Deadpan, Haunted and more — that play over whoever is on stage rather than replacing them, any number at once. Infatuated over Vessa is not the same person as Infatuated over Kova, which is the whole reason it works this way rather than shipping thirty-six presets.
Two of the four arrive with a mood their author dialled in for them. Those play underneath whatever you add, the board names them rather than showing them pressed, and clearing your own picks does not clear theirs — “Vessa is written hype” and “I have put this person in a hype mood” are two different statements, and only the second is yours to take back.
Their voice is theirs
The local speech service clones from an enrolled reference clip, and each companion names their own default in their manifest — so Vessa and Lucia do not sound like each other, and neither of them sounds like the app’s fallback.
The choice is kept per companion per speech model rather than as one setting for everybody, because the names that exist at all belong to whichever model is loaded. Swap the companion and the picker follows, instead of leaving behind a name the service has never heard of.
Two of them, on stage together
Two companions can share the stage and look at each other, take turns, and react to what the other one just said. Which is the point at which a desktop character stops being a widget.
More of them, later
The cast grows two ways: through the Workshop, where the folder format means an avatar is a directory anybody can publish rather than a build of the app, and through character packs. Everything on this page — the wardrobe, the voice, the written character, the moods — is available to anyone making one.
Pick one and see who you end up talking to.
Whoever you start with, they will be the one who remembers the conversation — memories are private to the companion they were told to unless they are the kind everybody should know.