Video summary

Can AI Make You a Top 1% Copywriter? I tested it

Main summary

Key takeaways

Business

Business-focused summary (AI-assisted copywriting workflow)

The speaker argues AI shouldn’t be treated as a “copywriter” that produces final copy from vague prompts. Instead, use AI to accelerate the pre-writing process—research, brainstorming, structure, and outlining—while the human provides the voice, storytelling, and final rewrite.


Core strategy & playbooks (as presented)

Secret 1: Stop obsessing over prompts; fix the inputs first

  • Problem: People blame “bad prompts” when outputs are generic.
  • Claim: LLMs don’t understand or strategize—they predict patterns.
  • Therefore: You must supply the right “patterns” by improving preparation.

Key principles

  • Lower expectations for what AI can do “by itself.”
  • Raise standards on the inputs/training before prompting (“garbage in, garbage out”).

Secret 2: Train AI with a “customer codeex” (not a static avatar)

Customer codeex = a detailed breakdown of the same customer across their journey stages, including:

  • fears, desires, and goals at each stage
  • changing objections
  • changing language and psychology over time

Illustrative example (customer “Sarah”)

  • Stage 1: ~$30K/month; business messy; small list; thinks email isn’t a priority; replies after months of email outreach
  • Stage 2 (12 months later): ~$100K/month; running ads; list ~20,000; new objection: “How do I know you’ll be better than me? I tried others and they didn’t sound like me.”

How it’s used operationally with AI Upload at the start of each chat:

  • customer codeex
  • sample of copy you like (style reference)
  • sample of the client’s voice (voice matching)

Result claimed

  • “Night and day” improvement versus single-shot prompts
  • Still not publish-ready without human finalization

Secret 3: Use proven persuasion frameworks to structure output

Claim: LLMs won’t reliably create persuasion structure unless you provide a blueprint.

Framework 1 (short-form): PAS

  • PAS is presented as workable for short-form copy (e.g., emails/ads) where the goal is a next-step action.

Framework 2 (long-form, primary): P.A.S.T.O.R.E (Ray Edwards “Pasta” variant)

  • P = Person + Problem/Pain
  • A = Amplify Cost of not fixing + Aspiration (what they want)
  • S = Story + Solution + System (how you discovered it + how it works for others)
  • T = Transformation + Testimony (proof it works)
  • O = Offer (what it includes, price, guarantee, etc.)
  • R = Request Response (CTA)

Framework 3 (prompt clarity): R.A.C.E

Used to reduce back-and-forth by defining prompt variables:

  • R = Role (who AI should be)
  • A = Action (what to produce)
  • C = Context (inputs/background)
  • E = Expectations (format, length, tone, constraints)

Alternative persuasion/prompt framework: CRAFT (Brian Albert)

  • C = Context
  • R = Role
  • A = Action
  • F = Format
  • T = Target audience

Self-improvement “bonus hack” (quality gate)

  • Add: “Score this output from 0 to 100 for persuasiveness/clarity.”
  • If score < 95, rewrite and score again.
  • Accept only when ≥ 95 (AI “competes against itself”).

Secret 4: Speed workflow (4 workflow hacks)

Hack 1: Don’t start new chats for the same project/client

  • Create a dedicated workspace (e.g., “projects”) per client/offer.
  • Upload context once:
    • customer codeex
    • brand voice guidelines
    • offer details + pricing
    • samples of preferred copy and the client’s voice
  • Benefit: compounding context; avoids re-pasting long instructions.

Hack 2: Prompt chaining (do tasks in steps)

Instead of a single giant prompt, break work into sequential steps that feed each other.

Example chain: market research

  1. Ask for a high-level market overview
  2. Pull exact audience quotes from forums (e.g., Reddit/Facebook groups/forums)
  3. Cluster quotes into buckets (fears/desires/objections)
  4. Map insights to awareness levels (based on Eugene Schwartz staged awareness)
  5. Connect insights to the offer by mapping features → why they matter to each pain point

Time improvement claimed

  • Market research: hours/days → ~30 minutes (speaker claims ~2 days to ~30 minutes)

Hack 3: “Handoff” across models (use the right model for the task)

  • ChatGPT: strong at research/structure/organizing information
  • Claude: claimed to produce more “human-sounding” prose
  • Credit-based model (speaker mentions “Manis”): better at reading large documents without summarizing away details (“reads every word” due to credit mechanics)

Operational example (objection extraction)

  • Upload last 100 sales call recordings → extract top 5 objections
  • Pre-address objections in ads and sales pages
  • Anecdotal outcome: increased conversions for a portfolio company’s VSLs (speaker’s “Vegas” story); specific % unclear, but implies “doubled conversion rates.”

Hack 4: Keyboard shortcuts for repeated prompt templates

  • Create shortcuts to insert frequently used prompts (e.g., “P1”).
  • Also recommends general OS shortcuts to improve efficiency.

Metrics / KPIs mentioned (and targets)

Outcome/impact claims

  • Helped 83 copywriters reach six or seven figure income
  • Generated over $1B for clients

Pricing/capacity economics examples

  • Student “Josh”: quoted $5,000 for a sales page; client wanted to pay $10,000
  • Student “Sam”: wrote 50 advertorials, made client over $50M; paid ~$70K total

Quality gate

  • Self-score must reach 95/100 before accepting output.

Time KPI

  • Market research improved to about 30 minutes (from hours/“2 days”).

Asset/voice constraints (examples)

  • Email length expectation example: under 150 words
  • Welcome email expectation example: ~300 words

Actionable recommendations (what to do next)

  • Don’t treat AI like an autonomous copywriter; treat it as an assistant that needs training inputs.
  • Build and reuse:
    • a customer codeex (journey-stage psychology and objections)
    • a preferred copy sample (style reference)
    • a client voice sample (voice matching)
  • Use structured frameworks:
    • P.A.S.T.O.R.E for long-form persuasion
    • R.A.C.E (or CRAFT) for precise prompt outcomes
  • Improve quality with a self-scoring loop:
    • require score ≥ 95, iterate if lower
  • Improve speed with workflow design:
    • dedicated project/workspace per client
    • prompt chaining for research → mapping → offer connections
    • handoff between models based on their strengths
    • keyboard shortcuts for prompt templates
  • (Business takeaway) Don’t just speed up—charge correctly:
    • speaker criticizes copywriters charging too low (e.g., $500) because it previously took longer
    • references a separate pricing video covering rates for different copy types (emails, sales pages, VSLs, ads, scripts, retainers, etc.)

Presenters / sources mentioned

  • Presenter: Sean Ferris
  • Referenced brands/companies: Forbes; Amazon; Netflix; Walmart; Tony Robbins; Alex Hormozi; ClickFunnels; Tai Lopez; Dan Kennedy; Dan Martell

Framework sources

  • Ray Edwards (Pasta / P.A.S.T.O.R.E)
  • Eugene Schwartz (awareness levels / staged awareness)
  • Brian Albert (CRAFT)

Models/tools referenced (high level)

  • ChatGPT (referred to as “Chat TBT”)
  • Claude
  • A credit-based model referred to as “Manis” / “Utari” (naming inconsistent due to subtitles/auto-generation)

Original video