I uploaded my resume into Cypher’s memory, then watched it fail to find a document it had already stored and indexed.

I traced the retrieval problem through two upgrades: source ranking, follow-up context, full-document expansion, and chunk truncation. Then I tested the exact conversation again.

Cypher found the resume, kept it active through the follow-up, and returned my complete work history.