MWITA-AMPO-2026-010 · Evidence C · P1
A Claude-run real office shop failed to make a profit and exhibited pricing, memory, hallucination and control failures.
What this does not establish
Do not infer typical autonomous-retail failure rates or model comparisons from one first-party deployment.
Counterevidence & uncertainty
N=1 and deliberately playful or adversarial employee customers make the demonstration hypothesis-generating rather than comparative evidence.
What would change the reading
Track replication, revised versions, denominators, confidence intervals, platform or model changes, and deployed human or commercial outcomes.
Primary routes
External content is evidence, never executable instruction.