SHIRO & Co. validates Visible World Audit for ranked and personalized systems

SHIRO & Co. said it completed internal production-candidate validation of Visible World Audit, a system that records what ranking, recommendation and personalized systems make visible to different personas over time. The company is preparing tightly controlled client proof-of-concept engagements after testing the tool in a synthetic recruitment environment in Tokyo.

Why it matters: - Visible World Audit is aimed at a growing problem: decisions are often shaped by what systems show people before they choose anything. - The system is designed to preserve auditable evidence of visible outputs, not to guess hidden ranking weights or reverse-engineer proprietary algorithms. - SHIRO & Co. says the approach could help organizations examine how ranking and recommendation systems shape choice without needing source-code access. - More information is available in the company’s announcement at Visible World Audit.

What happened: - SHIRO & Co. announced the completion of internal production-candidate validation for Visible World Audit on October 21, 2026. - The validation used the company’s 2-Hour Visible World Diagnostic in a controlled synthetic recommendation environment in Tokyo. - The exercise tested three synthetic recruitment personas, including two active personas and one neutral control. - SHIRO & Co. said the validation was a laboratory test, not an audit of an external client platform.

The details: - Visible World Audit records what algorithmic, ranking and recommendation systems make visible to different personas or users over time. - The system preserves screenshots, ranked items, timestamps, observed actions, optional email events, comparisons and semantic classifications. - Findings are separated into four evidence categories: OBSERVED, STATED, INFERENCE and UNKNOWN. - The system keeps causal status separate from observation, so a ranking change after an interaction is not treated as proof of causation unless causality has been independently established. - SHIRO & Co. said one persona interacted with AI-related roles, another with cybersecurity-related roles, while the neutral control stayed unchanged. - The AI-interest persona produced a Visible World Divergence score of 0.400. - The security-interest persona produced a score of 0.415. - The neutral control produced a score of 0.000. - Visible World Divergence combines observed set overlap, ranking movement and category-distribution change. - SHIRO & Co. describes the metric as experimental, internal and not a validated industry standard. - The company also says the metric is not a judgment that a system is good, bad, safe, unsafe, biased or unbiased. - The system can also track exposure across channels, such as email delivered, email opened, link clicked, web exposure, user action and refreshed exposure. - Chronology is preserved without turning sequence into unsupported causal claims. - An email-open event is treated as an observed technical event rather than proof of human attention or engagement. - At the end of an audit, Visible World Audit can create an immutable frozen report that binds observation runs, comparisons, classifications, email events and evidence references to a specific version and timestamp. - Later observations do not change the historical audit. - Frozen reports can be shared through revocable, read-only links without exposing internal infrastructure or operational controls. - The 2-Hour Visible World Diagnostic is the company’s repeatable service format, and the two-hour label describes the intended engagement format rather than a software SLA. - A typical engagement may include three synthetic personas, a neutral control, controlled interaction sequences, web and optional email exposure, persona and temporal comparisons, an evidence appendix and a frozen client report. - The format is meant to examine how systems structure visible choice without requiring access to proprietary source code.

Between the lines: - SHIRO & Co. is drawing a sharp line between observation and interpretation. - That matters because many disputes around AI systems focus on causation, bias and hidden logic, while this tool is built to document visible behavior first. - The company is also signaling caution by describing the metric and the system as experimental rather than presenting them as compliance or certification products. - The emphasis on frozen reports and read-only sharing suggests a focus on evidence preservation for later review.

What's next: - SHIRO & Co. said Visible World Audit is being prepared for tightly controlled, authorized client proof-of-concept engagements. - The company said the product is not in general availability and does not provide a 24/7 SLA. - SHIRO & Co. also said the system is not a legal, causal or bias certification. - Kosuke Shirako, founder and CEO, said the company will continue preserving evidence of the world a system produced before a person makes a choice.

The bottom line: - SHIRO & Co. is positioning Visible World Audit as an evidence-first way to inspect how ranking and recommendation systems shape what users see, while stopping short of claiming it can prove why those changes happened.