Retire old knowledge without erasing why a decision was made
Keep outdated documents out of current AI answers while preserving the historical context needed to understand earlier decisions.
Deleting every old document loses useful history. Leaving every version equally prominent makes current work harder. Teams need a clear way to retire knowledge while preserving its meaning.
Use status and relationships to separate current guidance from historical evidence.
Make retirement explicit
When a record stops applying, mark it as superseded or archived. State when it stopped applying, why and where the replacement lives. Put that notice where readers and retrieval systems can encounter it before the old guidance.
If there is no replacement, say so. An obsolete process should not remain the default simply because nobody has written the new one yet.
Preserve the reason
Keep the decision rationale and relevant historical scope when retention is appropriate. A team investigating an old customer experience may need to know which policy existed then.
That is different from presenting the old policy as guidance for a new customer. Ask assistants to distinguish historical questions from current operational questions and to surface the date and status of the source they use.
Update the paths that lead to it
Find active templates, project briefs and internal links that still point to the retired record. Redirect or update them where appropriate. Otherwise the team will keep reintroducing obsolete guidance into new tasks.
Review the actual search and retrieval behaviour too. A status field that the retrieval system ignores may not prevent stale material from influencing an answer.
Start with a small cleanup of the records most often cited incorrectly. Track whether those mistakes recur. The purpose of archiving is to improve what people and agents use today while keeping enough context to explain yesterday's decisions.