How I’d Make Money with Claude AI in 2026 (If I Had to Start Over)

Business goal (implied)

Create and sell a proven digital product (Gumroad) using Claude AI to rebuild/upgrade it, then validate distribution with paid ads first, followed by organic promotion if needed.

---

Step-by-step playbook (framework/process)

1) Find a product that’s already working (market validation)

• Use Facebook Ads Library to identify digital products actively advertising.
• Target a top-tier country (example: Germany).

• Use digital-product keywords and filter for ads running more than 5 days (rationale: testing takes ~3 days; if it’s still running, it likely performed).

• Build a short list in a Google Sheet with:

  • product name
  • landing page link
  • Facebook ad link

Concrete examples discovered

• “exercise book to improve your cello skills”
• A repeatedly seen niche: weight loss ebooks for men and women
• Chosen recreation example: “40-page hormone mapping protocol” (40 pages)

Selection criteria (product picking)

Pick the option with the “most potential,” based on:

• search demand
• audience size
• additional factors (not enumerated, but included in Claude’s analysis)

AI research step

• Upload the product list to Claude AI
• Use a prompt to rank/select the best option

---

2) Rebuild + “upgrade” the product using an AI content pipeline

• Don’t copy—upgrade and make the product unique.
• Connect Claude AI to Higgsfield AI to create a “monster” content studio:
  • Higgsfield: image/video generation
  • Claude: strong PDF/document generation with structured tables/charts/layout

Tooling / integration process (actionable)

Higgsfield

• Click MCP
• Copy the provided URL

Claude

• Go to Customize → Connectors → + (Custom connector)
• Paste the URL
• Name the connector
• Allow the connection

Generated deliverables

• Content is generated in HTML first (to look like a landing page), then converted to PDF later.
• Prompts (by step number) generate:
  • product positioning/name/description (saved for later prompts)
  • table of contents
  • chapters (one-by-one)
  • images per chapter (example: 2 images per chapter)
  • 3 cover variations (choose one; example: “cover A”)
  • final merge into a downloadable ebook

Operational note

• If Claude stops due to usage limits: wait ~4 hours and continue in the same chat.
• Generation time mentioned: ~25 minutes to get the download link after packaging/merge.

---

3) Package + publish on Gumroad (pricing + conversion tactics)

Gumroad setup steps

• Create/clear account view:
  • log out
  • make a new account (to show clean product analytics)
• Settings:
  • add brand name/logo
  • set theme (example: dark theme)
• Payments:
  • connect bank account
• Products:
  • create New product

AI-assisted Gumroad page creation

• Upload the ebook cover to trigger Claude to generate:
  • high-quality Gumroad cover
  • title
  • description + positioning (buyer-facing summary)

Pricing + discount mechanics

• Base price example: $49.99
• Add a coupon code (example: 25% off)
• After discount code:
  • price becomes $37 (implied: $49 × 0.75 logic)
• Configure the checkout form so the discount is offered only when available

Conversion belief / tactic

• Buyers usually decide based on the ad video
• Avoid complex landing pages

---

4) Drive first sales (distribution validation + scaling plan)

4A) Paid ads test (Facebook Ads)

AI creates ad creatives + targeting

• Prompt 7: Claude connects to Higgsfield Marketing Studio
  • generate 3 ad creatives (video ad scripts → final videos)
• Then generate:
  • exact audience targeting
  • ad copy/headline/creative guidance

Ad creative formats

• Example includes scripts/voiceover style messaging (e.g., hormone-related, “wrong numbers,” “4am,” “5 days at the gym,” etc.)
• Additional option mentioned: image ads (but only one image example used)

Campaign testing structure (explicit testing strategy)

• “One campaign, two ad sets”:
  1. ad set with interests
  2. ad set without interests
• Use 3–5 creatives duplicated across both ad sets
• Pixel tracking:
  • use a Facebook pixel to track sales
  • example mentions pasting pixel ID (“database” and “create it easily” are referenced)

Budget + timeline

• Suggested minimum: $10/day
• Example used: $15/day
• Test duration: 3 days
• Start time: “next day”

Interest build guidance

• Use Claude’s recommended interests
• Add what you can find in Facebook until audience size is “reasonable”

Example test results + KPI math

• Ad spend: $60
• Revenue returned: $124
• Profit estimate: ~$64
• Note: one buyer didn’t use the coupon code

Performance outcome:

• Non-interest ad set: “burning money” with zero results
• Interest ad set: generated 3 sales with “great cost per action” (CPA referenced qualitatively)

---

4B) Scale plan after finding a winner

• Let the test finish fully (even if they paused early)
• Scale the winning interest ad set
• Run country tests next:
  • example: 5 ad sets, each with one country per ad set

---

4C) Free/organic promotion method (short-form content)

Organic channel tactics (playbook)

• Use theme pages with link-in-bio on Instagram/TikTok
• Recreate best-performing faceless short-form video patterns
• Posting frequency:
  • 2 to 5 videos/day
• Key rule:
  • short-form platforms reward consistency
• Risk control:
  • “You only need one successful video to blow up your account—keep posting.”

Concrete creative production workflow

• Find free stock clips of women over 40
• Claude generates quotes tied to the ebook:
  • “help them” or emotionally hook them
• Edit video:
  • add clean effects
  • overlay quote so it stands out
  • add trending sound
• Post daily themed quote videos aligned to the ebook angle

---

Key metrics / KPIs & targets mentioned

• Testing filter target: ads running > 5 days (performance proxy)
• Ad test budget target: $10/day minimum (example: $15/day)
• Ad test duration: 3 days
• Example paid results:
  • spend $60
  • return $124
  • profit estimate ~$64
  • interest ad set: 3 sales
  • non-interest ad set: 0 results
• Organic posting target: 2–5 videos per day
• Scale target (next test): 5 ad sets for country testing (one country per ad set)

(No explicit CAC/LTV/churn/CVR targets were provided; CPA was only described qualitatively.)

---

Actionable recommendations (condensed)

• Use Facebook Ads Library with a run-time filter (>5 days) to avoid “random product” building.
• Create a product list (sheet) and let Claude rank potential rather than guessing.
• Build a full AI content pipeline (Claude + Higgsfield) to generate:
  • TOC, chapters, images, covers, and final ebook.
• On Gumroad:
  • publish quickly with a strong AI-generated description,
  • use a coupon discount to reduce friction (example: $49.99 → ~$37).
• For ads:
  • run interest vs non-interest ad sets,
  • use multiple creatives per test,
  • start at $10–$15/day and validate over 3 days.
• If interest wins:
  • scale the winning ad set,
  • then test countries with multiple ad sets.

---

Presenters / sources

• Presenter: The video creator (no name provided in subtitles)
• Tools / sources referenced: Gumroad, Facebook Ads Library, Claude AI, Higgsfield AI (via “MCP” connector), Facebook Ads Manager, Instagram, TikTok