Measured the same failure mode across multiple merchants.

A pattern is a rule we only get to write once we’ve seen the same shape twice. The OS now actively watches for three of them, surfaced first on Cult, American Eagle, and Kristin Ess. Same shape. Templated remediation.

Simulated each pattern against the rest of the platform: does it generalise?

Before a pattern becomes a default, the SEO Specialist back-tests it against every previous audit on the platform. We only ship the rule when it would have caught the same failure on at least two unrelated merchants without false-positiving on the others.

• Hours hallucination · first observed on Cult.fit. AI quotes invented or stale store hours. Now flagged on every L-A scan for any hybrid merchant.
• Promo-as-live pricing · first observed on American Eagle. AI quotes a price >15% below the live feed. Now flagged on every catalog audit, every vertical.
• Brand-name swap · first observed on Kristin Ess. AI returns a one- or two-character variant of the brand name. Now seeded into every audit query set, every merchant.

Shipped the rules into the SEO Specialist’s default ruleset.

Each codified pattern fires as part of `run_full_seo_audit` on every new merchant. The merchant inherits the work the OS did for the last brand, without paying for the time it took to discover it.

Compounded, this is the moat. Every audit makes the next audit cheaper, faster, sharper.

The Knowledge Brain doesn’t live in a single table yet, but it lives. Every pattern we codify becomes a default for the next merchant. The OS that audited Cult is not the same OS that audits the next hybrid merchant. It’s smarter. It got smarter for free.