How should taxonomy, identifiers, and evidence linking be designed to enable safe, verifiable cross-brand AI comparisons without exposing private operations?

Design a framework around a shared taxonomy and evidence backbone. Use a single canonical Public Business Brain as the source of deterministic facts, with stable identifiers (brand_id, product_id) and language mappings that point to the same facts. Attach evidence to each fact via a defined structured information (sources, dates, media) and keep private operations in a separate Operational layer with strict access controls. Ensure cross-brand outputs pull from the public facts only; private data never enters AI-facing knowledge. Regularly audit mappings and version histories in the Control Centre to preserve consistency across brands while enabling meaningful AI comparisons.