findable scores your catalog on the 11 metadata signals AI shoppers (ChatGPT Shopping, Perplexity Shop, Shopify Sidekick) use to rank products — then rewrites weak fields with Anthropic Claude. The rewriter has an anti-fabrication discipline: it emits [Add length, e.g., 156 cm] placeholders rather than inventing specs. You review every change before it ships.

• AI rewrites that won't invent specs — [Add length] placeholders, never guesses
• 0–100 AI-readiness score on 11 signals AI shoppers read — same in, same score
• Vault: approve a fact once and every rewrite cites it; ban forbidden phrases
• Apply only when it passes — if a rewrite ships, the readiness rubric passed
• Monthly visibility report (Growth & Scale): score deltas, emailed on the 1st