What we are trying to learn

The core question

Can important work keep moving across many AI conversations — with free discussion, optional keeping, finding originals and deliberate changes of view — without asking people to maintain structure after every exchange?

What this website shows

Prepared, synthetic examples you can operate to understand the design and its boundaries. They do not show the quality of a real AI, whether people understand the design without help, better decisions, retention, willingness to pay or market demand. Engineering checks confirm that specific behaviour works; they are not research with people.

How we separate claims

We keep apart what was observed, how it was interpreted, judgments we have adopted, and results that have been shown.

Questions still to test

Do people understand which material, and which version, they chose?

Can they come back, continue and find the original?

Can they tell a suggestion from something they confirmed?

Can a recipient understand a brief on its own?

Published material

There is no public research material yet. Nothing on this page reports a study with people.