Ommi builds a personal AI clone — one that answers the way its owner would, in their tone and even their voice. The hard part isn’t the model. It’s getting a person to describe themselves well enough for a clone to be believable, without it feeling like a questionnaire.
So setup became a conversation in four steps — Purpose, Identity, Taboos, Insights — each with a time estimate and templates to start from. After every step the owner sees how their clone would answer right now, so each answer visibly changes something. The voice comes from a short script read aloud, and sources — notes, messages, decks from Drive, Notion or Slack — teach the clone the rest.
The brand grew from the same idea of a digital self: a wordmark over a lit silhouette, and a pixel-art hand for every template — a human gesture, rendered in pixels.
The brand and the product were designed in eight weeks, and Ommi has launched its MVP.
