Real build: POST to HubSpot Forms API on submit; fire Amplitude events where stubbed.

Open items carried into this build (flagged rather than invented):

• rmm_four_question_result cannot be computed honestly from the RMI set. The Four-Question Test is "good or bad / by how much / why / what next"; the six RMI questions only cover the last two (Q1 maps to "why", Q4 to "what next"). This build emits a four-character string with "-" in the two positions nobody is asked. Lucija to confirm the mapping, or drop the field.
• Proof block: brief section 6 says role changes the proof block, so the role proof is what renders. The level-generic quotes in copy-deck section 5 are held as a fallback. Worth confirming: it means most of those quotes never appear on screen.
• L1 result-page hook: the before-and-after visual from the beginner's guide is a placeholder box. The image still needs exporting; the guide itself is deliberately not linked (copy deck section 5).
• Consultant / partner path: no source messaging exists (copy deck section 8, item 1). The CTA links to the partner program and the sub-line is build copy, so it needs the partner owner's review.
• L3 "departmental rollout guide" secondary: neither the asset nor a form variant exists (copy deck section 9). Rendered with a one-field email capture so the flow is testable. Build it or drop it.
• L4 resources strip is empty by design: Zebra AI copy is held until the discovery period closes (copy deck section 8, item 2).
• Starter-kit consent: a marketing-consent checkbox is added, because the whole offer is an email sequence. Not specified in section 7; confirm with lifecycle marketing.
• Resources strips at L1b and L1c are picked from the D16 asset pool (product tour, guides). The deck names only the secondary for those branches.

Event log (this session):

- nothing yet -