Skip to module content
🧠AI Academy
0/91
Sign in
Module 21 · ~5 min

Demonstrable Proof

What you must be able to show live — the demo bench.

Reading progress
0/8 · 0%

The big idea

💡Key idea
The AI-agency market is full of slide decks. Close rates follow live demos. Maintain a bench of six demos, each rehearsed, each timed, each with a recorded fallback: (1) the 10-minute audit — run on the prospect's domain during the first call; (2) the 60-second skills library tour — shows the Google Ads suite firing on a sandbox; (3) the research flex — live multi-agent market map on their category; (4) the content chain — their last post atomized into a full set via Canva MCP; (5) the citation check — their brand vs competitors across five money-prompts in ChatGPT/Perplexity/Claude; (6) the safety story — permission matrix plus a hook that blocks an unauthorized mutation attempt. The sixth demo closes more deals than all the capability demos combined.
Quick check
1 question · instant feedback
0/1
  1. The 6th demo (safety story) matters because:

Numbers that matter

6
Demos on the bench, minimum.
60s
Skills-library tour target — the 60 seconds that explains the entire company.
Quick check
1 question · instant feedback
0/1
  1. Recorded fallbacks are:

Deep dive

2/2 open

Each demo targets a different buyer concern. Together, they cover the full range of what a sophisticated buyer needs to see.

**1. The 10-minute audit.** A single skill running live during the first call. You take the prospect's domain, run the GEO access-check and extractability score, pull the tech stack, and surface three automation hypotheses — all while they're on the call. This is the wedge.

**2. The 60-second skills library tour.** The Google Ads suite firing on a sandbox: pull → diagnosis → dry-run push → approval gate → audit log. Sixty seconds explains the entire company model.

**3. The research flex.** A live multi-agent market map on their category. Show the orchestrator dispatching subagents; show the synthesis. Makes the capability concrete and relevant.

**4. The content chain.** Their last blog post → atomized post set plus a carousel built via Canva MCP, in their draft voice. Specific to their brand, built in real time.

**5. The citation check.** Their brand versus competitors across ChatGPT, Perplexity, and Claude for five money-prompts. Shows GEO visibility (or the lack of it) in a format that's immediately legible to any marketer.

**6. The safety story.** The permission matrix plus a live demonstration of a hook blocking an unauthorized mutation attempt. This is the one that closes.

Every AI agency can demo capability. Show a generated email, a summarized report, an automated workflow — it's table stakes.

Almost none of them can demo structural refusal.

When a buyer watches a hook block an unauthorized mutation attempt in real time, the message is clear: this team didn't just build something that works. They built something that knows its limits, enforced in code.

That's the trust deposit that closes. A buyer who's worried about AI going wrong — spending money it shouldn't, sending emails it wasn't supposed to — isn't reassured by another capability demo. They're reassured by watching the guardrail work.

Quick check
1 question · instant feedback
0/1
  1. The skills-library tour target length is:

How to run it

  1. Ship all 6 demos
    Timed, rehearsed, recorded fallbacks in place.
  2. Under 60s for the skills tour
    The single most persuasive artifact.
  3. Always include the safety demo
    Refusal beats capability.
  4. Recorded fallbacks
    Network/API failures don't kill the pitch.
Quick check
1 question · instant feedback
0/1
  1. The 10-minute audit runs:

In the field

🔬Worked example
Demonstrated refusal builds more trust than demonstrated capability. The safety-story demo — showing a hook that BLOCKS an unauthorized mutation attempt — closes more deals than the flashiest capability demo.
🚫When not to reach for it
Never demo without recorded fallbacks. Live demos fail; recorded fallbacks are the difference between 'ambitious' and 'unprofessional.'
Quick check
1 question · instant feedback
0/1
  1. Slideware vs live demo:

Pitfalls & takeaways

Failure modes

  • Slideware instead of live demo.
  • No recorded fallbacks for network/API failures.
  • Long demos — the 60-second skills tour outperforms 20-minute walkthroughs.
  • Demoing only capability, never refusal.

Durable takeaways

  • Close rates follow live demos, not slideware.
  • Demonstrated refusal beats demonstrated capability.
  • Recorded fallbacks are the difference between ambitious and unprofessional.
  • The 60-second skills tour is the most persuasive artifact you own.

Do the work

🏋️Prove you learned it

Build the 10-minute audit demo end-to-end. Take a real prospect domain (or your own site), run the skill, produce the deliverable, and time it. Record a fallback video. Ship a one-page timing card for each of the 6 demos.

0 chars
📦Artifact to produce
6-item demo bench + recorded fallbacks + a per-demo timing card.

Sources

  • · GMS Field Manual §21 (Demonstrable Proof)