• Setup (30 seconds): open the Pilot, click ▶ SCRIPTED DEMO, confirm the SCRIPT panel is on (🎤 SCRIPT button, bottom bar). Each scene shows your narration in the gold panel — glance, don't read.
• The one idea to land: the numbers running every restaurant are unverified estimates; MGE is the independent, deterministic engine that certifies them — and every claim in this demo is falsifiable on the spot.
• Posture: never say “imagine if” or “this would.” Everything on screen is running. When in doubt, hand them the RUN button.
• Framing line to open: “This is an eight-minute story in six scenes. Interrupt me anywhere — everything I show you, you can run yourself.”

• “This is a canned demo.” — “Upload your own payroll or invoice file — the engine analyzes it live and the results change. A scripted demo can't do that.”
• “Any BI tool could build this.” — “They can build the calculation; they can't certify it. R365 certifying R365 data is worthless — same reason companies don't audit themselves. Independence is the architecture, not a feature.”
• “Who decided the Trust Score weights?” — “It's a deterministic weighted sum — completeness 30, consistency 30, timeliness 20, alignment 20. No model, no inference. Run it twice, same score.”
• “No revenue yet.” — “Right question is: does the engine run, and is the problem structural? $28K–$84K per 10-unit operator per year says yes. Revenue follows certification volume; volume follows trust in the output.”
• “Too narrow — restaurants only.” — “65% of MGE is vertical-agnostic CORE. The 198 rules are the restaurant instance; swap the instance layer and the same CORE certifies hospitality procurement or healthcare billing.”

End on the Share tab: export the After-Test Report and hand them the PDF — their run, narrated: what they saw, what MGE did, what it proved, why it matters. Then the closing line: