01
Source before inference
A generative model is only as meaningful as the human material behind it. The archive comes first, and the model stays downstream of it.
Creomind Principles
Standards we intend to be judged against.
A Cognitive Model is built from the most personal material a family will ever hand over. These are the rules we build against, written plainly enough that anyone can hold us to them.
02
Archive and Generative remain distinct
What someone said and what a model infers they would say are different forms of knowledge. They are labelled differently, presented differently, and never merged.
03
Uncertainty must be visible
When evidence is limited, Creomind makes that limitation visible. It may decline to answer or, where the user chooses to continue, clearly identify the response as speculative. Confidence is described in plain language, never as an invented score.
04
A human is allowed to contradict themselves
People hold opposing beliefs, and change their minds. The Cognitive Model represents that complexity rather than averaging it away.
05
Generated answers stay connected to evidence
Where evidence exists, a generated response remains traceable to the memories, recordings and documents that shaped it, with dates and sources a family can check.
06
The person retains agency
The subject decides what is captured, what is kept and who may access it. Source material and the archive belong to the person who made them and the family they choose. Export is in open formats, and consent is revocable.
07
Presence without deception
A generated answer may speak in the first person and in a modeled voice, because that is how a conversation feels human. The interface still marks it as generated everywhere it appears, including in the export. Presence on the surface, provenance underneath.
08
Technology should serve the person, not overshadow them
If a visitor remembers the software more than the voice, the design has failed.
09
The model should know its limits
Knowing where the evidence runs out is a feature, not a failure. A Cognitive Model is built to say when it does not know enough.
10
A person is more than a chatbot
A conversational surface is one way into a Cognitive Model. It is not the person, and we will not present it as one.
If another company copies the product, we would like them to have to answer these questions too.
Start your archive while you can still speak for yourself.
One conversation is enough to begin. We work with a small number of people at a time, and we discuss cost plainly before anything is decided.
Living subjects only · Consent first · Reply within 48 hours