How knowledge gets adopted

• Import — hand over what you have: your website, YouTube videos, recordings, conversations with the AI.
• Extract — the AI finds knowledge candidates in it. Each candidate carries the source and the spot it came from.
• Adopt — a person reads each one and adopts it, passes on it, or uses it to update knowledge that already exists.

The four angles below are what a person checks when adopting. They are not judged automatically by the system.

What in the business is this knowledge for? Knowledge with no purpose gives the AI no way to tell when to use it. If you cannot find what it serves, pass on it or reconsider where it belongs.

What is the knowledge based on? Extracted knowledge keeps its source and location, so its evidence can be traced even after adoption. Whether the evidence is out of date, or rests on a single source, a person checks from the sources shown on screen.

Knowledge connects through typed relations:

• extends (adds detail) / depends on / is an example of
• constrains / overrides / supersedes (is a newer version)
• contradicts

When similar knowledge already exists, the two texts appear side by side, so nothing is stored twice. They are merged only when a person approves. When two pieces of knowledge contradict each other, the AI does not decide which is right; it lays the conflict out for a person to judge.

Does the knowledge belong to the whole company, to a brand, or to a site? Too wide a scope binds other brands; too narrow a scope leaves gaps where it should apply.

Edit a piece of knowledge, and the plans and per-channel drafts that use it are marked for review. An explanation connected to an external data source is also marked when that data is updated. Knowledge is not adopted once and forgotten; it is re-checked as it is used.