40 — Demo Script
Status: CANONICAL · Commercial OS
Owner: Founder (sales)
Last verified: 2026-07-21
Performance notes: [BRACKETS] = personalize from discovery. (italics) = stage directions. Beats fixed per 39 Demo Guide §2.
Purpose
The word-for-word 15-minute demo. Deliver it enough times that you can leave it — but the beats, the climax, and the refusal never move.
Beat 1 — Their Morning, Named (0:00–2:00)
"Before I show anything — [NAME], you told me something I wrote down word for word: '[PAIN SENTENCE from discovery]'. And the morning behind it: [CRM], then [SUPPORT TOOL], then [BILLING], then [CHAT] — [TOOL COUNT] tools to answer one question: what's true about this account today? About [H] hours a day of that, per person.
And the AI part — your team tried it. Now it writes emails. Not because the models are bad — because the AI forgot what they told it, stopped halfway, and made things up.
For the next 12 minutes I'm going to show you the same morning with all of that gone. One rule for me: everything I show you is real product, live. If you ask about something we haven't built, I'll say 'not yet.'"
(Screen share opens on the workbench — already loaded. No app tour, no settings, no slides.)
Beat 2 — The Workbench Opens (2:00–5:00)
"This is a Customer Success workbench. Not a dashboard — watch what it opens on. Three things, every morning:
What changed. Overnight, across every connected tool: [scroll] — Meridian Corp's product usage dropped 42% over three weeks. Notice nothing's asking you to go check four systems — the checking already happened.
What needs you. Meridian's renewal is in 45 days. And their champion — [contact] — hasn't replied in 19 days. Those two facts live in different tools in your world. Here they're the same story.
What's prepared. Your 10 o'clock meeting brief — already drafted. We'll come back to who wrote it and how you'd trust it.
[NAME], honest question — this morning view versus your [TOOL COUNT]-tab version: what's the first thing your team would do with it?"
(Let them answer. This is the "that's my morning" moment — don't step on it.)
Beat 3 — Evidence Under Everything (5:00–7:00)
"Here's the part that makes this different from every AI tool your team demoted. Click the usage-drop signal — evidence. Which systems, which numbers, when: usage from [ANALYTICS], the support escalation from [SUPPORT TOOL], the renewal date from [CRM], and this line from an email thread — 'budget review next quarter.' Four tools, one story, every claim with a source.
Click any fact — (open provenance panel) — source system, timestamp, confidence. Your organization's truth, assembled, with receipts. We call where this lives the Spine — your operational truth, in your tenant, exportable any day you want. No AI wrote any of this. AI is not allowed to write truth here. That's architecture, not policy."
(P3/P4 in the room: 60 extra seconds on entity resolution + tenant isolation. Else move.)
Beat 4 — Work With the Twin (7:00–11:00) — THE CLIMAX
"Now the part your team stopped believing in. Every person here has a Twin — theirs, scoped to their role, working on the same bench. Not a chatbot in a sidebar. Watch the difference.
Ask: (type: 'What did we commit to Meridian in the March QBR?') — answer, with the commitment, who made it, where it's recorded. It didn't guess — it read your Spine. If it had nothing, it would say nothing. Silence is a feature.
Assign: (click: 'Draft renewal outreach reflecting the usage conversation') — the Twin's drafting from the actual account story: the usage dip, the March commitment, [CONTACT]'s last thread. Not a template with a name plugged in.
And now the moment that changes AI from a toy to a colleague. The draft's ready. Does it send? No. (pause) It's a proposal. Evidence attached, reasoning shown. [SKEPTIC NAME] — you're approving this. What would you check first?"
(Take their answer seriously — click into whatever they name. Then:)
"You'd edit this line — (edit it) — and approve. (click) Now it executes — through your actual [EMAIL/CRM], the systems you already trust. And there's the receipt: proposed, evidence, approved by you, executed, result. (3 seconds of silence — let the receipt land.)
The Twin prepares. The human approves. Every time. That's the whole philosophy in one click."
Beat 5 — The Refusal + The Trail (11:00–13:00)
"Two more things nobody else will show you.
First — what it won't do. (attempt an action without approval path / show a memory-write proposal) — the AI wants to remember 'Meridian is at risk of churn.' It cannot write that. It can only propose it. (show the TruthLayer queue) You approve it, it becomes organizational memory. You reject it, it never existed. Every AI you've used auto-writes its memory — it guesses about your business. Here, your organization's memory contains only what a human verified. Every AI remembers. Only IntegrateWise remembers the truth.
Second — the audit trail. (scroll it, slowly) Every proposal, approval, execution. I know it's boring. Boring is the point. This is the tab your security team reads, and it's why they'll say yes."
Beat 6 — The Ask (13:00–15:00)
"Remember the napkin from our first call: [H] hours × [N] people × $[C] — about $[MONTHLY VALUE] a month in reconstruction work. What you just watched removes most of it — and you saw how, not just that it does.
Here's the step I'd propose. Thirty days, up to ten of your people, your real tools — [THEIR STACK]. We agree on three success criteria up front — I'd suggest: morning prep under 10 minutes, your next QBR prepped in under 30, and at least one catch like the Meridian signal on a real account of yours. Day 28, we look at the numbers together and you decide. If the criteria aren't met, you'll know exactly why, from me first.
What would you want the third criterion to be?"
(The close is a question that starts the POV scoping — 37 Play 4 takes it from here.)
Variants
- 8-minute exec version (P5 present): Beats 1→2→4(compressed to Assign+Approve)→6, with the continuity line added: "Month 12: institutional memory that leaves with people no longer matters."
- Technical deep-dive (P3/P4 follow-up, 30 min): Beats 3+5 expanded — entity resolution, boundary contract, fail-closed demo, export walkthrough. Script: 21 Platform Story + 42 Security FAQ.
KPIs: 39 Demo Guide owns demo metrics. Script-fidelity spot checks on recordings — beats present, climax landed, refusal shown.
Related: 39 Demo Guide · 19 Product Narrative · 38 Discovery · 37 Playbook