From AI Demos to Defensible AI Operations: A Redacted Enterprise Problem Map for Replay, Provenance, and Claim-Bounded Agent Workflows

This v1.2.0 forward correction updates the published v1.1.0 working paper without altering itsfrozen record. It replaces stale pre-publication DOI and license wording; bounds claimsabout reproducibility, independence, public disclosure, and evidentiary effect; updates current Foldcross-links and bibliographic metadata; and removes an internal deposit checklist from the articlebody. The metric facts and the paper's R1-R3 claim ceiling are unchanged.The historical v1.1.0 public bundle remains unchanged. v1.2.0 does not republish its source or harness ZIP and includes only the rendered paper and public release documentation. The paper remains a working paper based on fixed synthetic fixtures: it is not an enterprise validation study, customer or deployment evidence, independent verification, legal analysis, or a conclusion about patentability, priority, or freedom to operate.