Video summary

This AI VSL Framework Made Me $5,000,000

Main summary

Key takeaways

Business

Business outcomes / claims

  • Creator reports writing VSLs (video sales letters) with AI for ~2 years, generating “close to eight figures” online across info/education companies.
  • Achievements referenced:
    • New “WAP plaque” for another $1M award (and previously $1M ClickFunnels plaque).
    • Prior agency validation: $4.5M in one week for a client (via VSL work).
    • Client copywriter pricing cited: $15K–$25K per VSL in an older workflow.
  • New workflow speed:
    • VSL creation: 4–6 weeks → 1–3 days
    • Revisions: weeks → 1–3 days
    • Enabled by AI plus prompt/psychology knowledge.

VSL definition & purpose (what it is operationally)

A VSL is described as a:

  • High-leverage scripted sales asset designed to sell one offer to one specific audience segment.
  • Built to drive specific action (e.g., book a call), break objections/limiting beliefs, and frame the offer as the solution.

Conversion framework (VSL structure / playbook)

The VSL “engine” follows a sequence (slightly varied by offer):

  1. Hook
  2. Pain
  3. Agitate
  4. Story
  5. Mechanism (how the solution works)
  6. Proof
  7. Offer
  8. Urgency
  9. CTA (call to action)

Note: the creator mentions a relation to the “hero’s journey” / story arc, but uses the above as the practical checklist.

Concrete example: AI-themed VSL (mechanics + messaging)

He maps his AI offer with the following components:

  • Hook: “The world is changing—get up to speed with AI or you’ll be stomped.”
  • Pain/Agitation (example data point):
    • Claims of job displacement, including an anecdote:
      • “Square released 60% of the workforce of ~10,000 employees due to AI.”
  • Story:
    • Took “short-term pain” by pausing revenue for ~1 year to build an AI company (Kendo).
  • Mechanism (two-part business model):
    1. AI freelancer / AI operator: “a couple clients + AI tools,” targeting $3K–$10K/month.
    2. AI insiders mastermind: for business owners / those running high-level AI agencies; claims of members doing millions/month.
  • Proof:
    • Running the approach for 2–2.5 years, plus testimonials/case studies.
  • Offer + CTA funnel:
    • Book a call for the main offer.
    • Run a lower-ticket webinar in the next couple weeks; webinar purchases are upsold to a higher tier.

Operational shift: “old vs new” production system

Old way (service/production-heavy)

  • Cost: $5K–$25K per VSL
  • Timeline: 4–6 weeks
  • Constraints: influencer/figurehead filming and more manual iteration.

New way (AI-assisted copy pipeline)

  • Timeline: 1–3 days
  • Core claim: Iterations happen via prompting; AI handles writing + research.
  • Humans provide:
    • psychology understanding
    • editing/quality control

Playbook: building a VSL with AI (end-to-end process)

Presented as a “pyramid” of inputs → output.

1) Research (foundation)

“99%” of failures come from bad starting research.

  • Use AI research tools (mentions Perplexity and Gemini).
  • Principles:
    • Back every claim with data (trust-building rule).
  • Two approaches:
    • AI-driven research from ICP + topic
    • Deep ICP research via an internal framework:
      • “~50 questions inside of AI Insiders”
      • Called the ICP framework that “one-shots” writing quality.

2) Build “context files” (what you feed the model)

Create structured inputs so the AI writes accurately and on-brand:

  • Offer document
    • What it is, price, target, not-for, promise/transformation, mechanism
  • Brand voice document
    • Gather raw creator speech (YouTube transcripts, sales calls)
    • Use tools like WhisperFlow to transcribe and extract voice profile
  • ICP research document converted to JSON
    • Reason: long PDFs may get summarized/skip details; JSON enables line-by-line parsing.

3) Set up a model project (segmented workspace per VSL)

  • Create a Claude project per VSL
  • Upload the context files into “project knowledge”
  • Add project instructions + priming prompt so it doesn’t “guess,” but uses your filled inputs.

4) Validate output before drafting

Have Claude confirm it understands:

  • client voice
  • real pain points
  • true objections (narrow, specific)
  • language lifted from ICP

5) Generate the VSL script using the structure

Additional writing specs:

  • Speak in brand voice
  • Write to one person (micro-targeting: “one person, not everybody”)
  • Style guidance:
    • short sentences, varied rhythm, conversational
    • no corporate buzzwords
    • open loops/curiosity
  • Target length:
    • 15–20 minutes spoken context
  • Add after each major section:
    • a brief psychological purpose (for review/learning)

Editing & quality control (human layer that makes it “not AI”)

He emphasizes that AI provides scaffolding; humans must produce authenticity and specificity.

Key editing targets

  • Remove “polished AI words” / delete generic filler
  • Break rhythm (spoken texture, punchy lines; include performance cues like pauses)
  • Replace abstractions with concrete specifics
    • Example shift: “achieve financial freedom” → “stop checking your bank account before buying groceries”
  • Inject client-specific stories and phrases (AI won’t know personal history)
  • Inject specific numbers
    • Make pain feel “real” without being insulting

Tone guidance for pain

  • Don’t frame as “me vs you” (attacking the prospect)
  • Frame as “us vs them”
    • enemy = the risk/market shift/approach that’s failing

Editing workflow

  • Read cold + read out loud
  • Sanity check: “Would a human say this?”
  • For pain/proof:
    • replace generic case details with specific case-study points
  • Use a feedback loop with Claude, e.g.:
    • “This feels generic—make it more conversational and specific”
    • “Brand voice is off—match the client’s tone”

Scoring hack (iterative optimization loop)

  • After multiple iterations:
    • upload the full draft back to Claude and ask it to:
      • score against copywriting principles (mentions Dan Kennedy, Russell Brunson, Jason Fladlien)
      • request changes needed to reach 100/100
  • Repeat until the score converges and it reads well out loud.

KPI / economics framework (how to price/justify VSL ROI)

He provides an example conversion math:

  • Assumption: VSL converts at 3%
  • Traffic: 1,000 visitors/month
  • Price point: $997 offer
  • Result:
    • ~$30K/month revenue from that VSL

He also notes sensitivity to higher conversion:

  • If conversion rises to 5%, income increases materially (direct lift via math).

VSL pricing models mentioned

Client deal structures:

  • Performance-based
    • take a “percentage of upside” if the VSL beats current performance after thresholds
    • suggests minimum 2,000 watches/views/play-throughs
  • Flat fee (he says he has sold this before)
  • Revenue-share / percentage-based ROI deals

Implied goal:

  • Guarantee better performance than the current VSL under measurable conditions.

High-level recommendations (actionable takeaways)

  • Treat VSL as a repeatable system:
    • Research → Context files (offer + brand voice + ICP JSON) → Claude project → Generate script → Human editing → Iterative scoring loop
  • Use one-person targeting (avoid “everyone” messaging).
  • Enforce a claims-with-proof rule for trust.
  • Edit for human texture:
    • rhythm, specificity, real stories, real numbers, not generic AI phrasing
  • Measure ROI via conversion rate lift and performance thresholds, then structure pricing around it.

Presenters / sources mentioned

  • Presenter: The unnamed creator/speaker throughout the subtitles
  • Company/brands referenced: ClickFunnels, Kendo, AI Insiders, Square (workforce reduction anecdote)
  • AI tools referenced: Claude, Perplexity, Gemini, WhisperFlow
  • Copywriting thought leaders referenced for scoring principles: Dan Kennedy, Russell Brunson, Jason Fladlien

Original video