Most collaboration friction comes from unstated preferences: how direct you want feedback, how much context you need before deciding, how you behave when unsure. The AI User Manual makes those explicit — written from observed interaction, not from a form you filled in.
The metaphor is deliberate. You are not a type to be classified; you are a person with working preferences. A manual says here is how to work with me well — which is more useful, and more respectful, than a label.
Communication directness and verbosity, how you prefer to receive disagreement, your decision style under uncertainty, how much context you need before committing, and what helps you feel understood.
A manual you cannot edit is just another label. Every line can be corrected, and corrections regenerate your model while preserving the previous version so you can see exactly what changed.
Q: Who can see my User Manual?
A: Only you, until you explicitly share it. Sharing generates a separate link; nothing is public by default.
Q: Is this used to profile me for advertising?
A: No. KKMatch does not infer sensitive attributes and does not sell behavioural profiles.
Q: How is it generated?
A: From a roughly five-minute conversation, plus — only with your consent — voice and video derived interaction features. It is structured output, not free-form guesswork.