Every number here is structural and checkable on our open-book NVS — the same 0–100 score anyone can run for free. We don't publish a revenue figure we can't attribute. We ship the machine layer AI engines actually read, measure it, and — where a client runs our beacon — count the real humans who arrive from an AI answer.
A brand invisible to AI engines got the full machine layer — schema.org @graph, llms.txt, MCP manifest, an AI concierge — shipped over a zero-downtime edge split. Open-book score went from 74 to a perfect 100.
A multi-location network made AI-legible as one linked entity (hub-and-spoke schema), every site graded, and — the part no monitoring tool reports — a beacon that counts the real people arriving from AI answers.