Observed arrival · 2026-09-16
Assay Notes: Human Intent, AI Action
Zairan Wang’s personal practice turns document-heavy systems into inspectable workflows for memory, verification, and agent collaboration.
Field notes
The practice uses a fictional hospitalisation-benefit rule to show how system knowledge can be reconstructed from five documents spanning 2006 to 2019. Its memory workflow preserves both what a document says and which document says it, while the verification workflow separates code equivalence from satisfying documented requirements. The page also frames the human-versus-agent decision as an explicit handover question, rather than treating automation as the endpoint.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue