How VGO works
VGO is an open action loop. The fictional walkthrough below shows how a team responds when a generative system states an obsolete product capability. It does not claim that any before/after observation proves causation.
Scenario: In a pre-stated buying-intent question, a generative answer describes a product capability the brand no longer offers.
1. Diagnose
Input：User intent scenario, full answer text, citations, surface, time and conditions; current first-party facts.
Action：Compare the machine statement with verified brand facts and note gaps, uncertainty and possible harm.
Output：A recorded perception gap with evidence and limits.
2. Decide
Input：Gap record, user value, expected usefulness, cost, risk and feasibility.
Action：Prioritize the issue, assign an owner and set a review point.
Output：A prioritized action with rationale and constraints.
3. Act
Input：Owned facts, related content and legitimate external corroboration paths.
Action：Update maintainable first-party sources and seek independent correction where appropriate.
Output：Published changes with owners and review conditions.
4. Verify
Input：Comparable question, surface and conditions; raw and contrary evidence.
Action：Retest under comparable conditions and check factual correctness and provenance.
Output：Comparable observations that separate change from outcome claims.
5. Learn
Input：Actions, results, failures and remaining limits.
Action：Keep reusable facts, evidence and decision records; revise priorities for the next cycle.
Output：Updated owned assets and learning for continuous improvement.
V1.0 public evaluation focus remains search and generative AI. Social search, recommendation and AI distribution are visibility surfaces within the machine-mediated world; they are not yet a separately validated public-evaluation scope.