This is the exact system from my Book of Enoch challenge: AI picked the product, built the store, ran the ads, and we were profitable on day one. Every prompt I used is below, generalized so you can run it for your own product. All you need is the Claude desktop app and a free account on Whop.
Step 01
Set up your AI stack
Two tools run this whole business. The Claude desktop app is the brain: it researches, decides, writes, and runs agents for different roles. Whop is the body: website, checkout, payments, pixel, ad tools, all with an API so Claude can operate it directly through the Whop CLI.
I chose Whop over Shopify for one reason: no monthly subscription eating the budget, and the API covers everything I need to run the business from inside Claude.
Set your rules before you start. Mine were: the AI makes every decision, all content gets generated with AI, and the total budget is $500 on a virtual card the agent controls. Rules force the interesting result.
Step 02
Let AI find your product
Don't invent a product. Give Claude tight constraints and let it hunt for existing demand. Mine came back with the Book of Enoch: a public-domain ancient text getting millions of views online, that anyone can legally sell as a beautiful hardcover.
Find me a physical product we can launch quickly with three characteristics:
1. There needs to already be demand — people are actively searching for it or content about it is getting millions of views right now.
2. I don't want to buy inventory upfront — it must work with print-on-demand or dropship fulfillment.
3. There needs to be a real reason we can compete — public domain content, an underserved angle, or a format nobody is doing well.
Do not invent a novelty product I'd have to convince people they need. Find something people already care about. Give me your top 3 options with the demand evidence, the fulfillment path, and the competitive angle for each.
Step 03
Start distribution on day 1
Before the product existed, we started building the audience. On a $500 budget, organic content on X is the cheapest shot. Claude named the brand (Enoch's Archive), generated the logo and banner, then built a posting machine.
We're launching [PRODUCT] to the [NICHE] audience on X.
1. Come up with a brand name and generate a logo and banner that fit the niche's aesthetic.
2. Research what posts in this niche are already getting high engagement — formats, hooks, topics, posting times.
3. Turn that into a posting strategy, then set up scheduled tasks so that every day the account posts 5 times, replies to 10 popular posts in the niche, and makes 2 quote tweets.
The goal is a theme page that earns steady organic traffic without ad spend.
Fair warning from my run: a brand-new account gets almost no reach for the first days. That's normal. Step 12 covers how we fixed the cold start.
Step 04
Make the product
Claude compiled the public-domain text and formatted it into a print-ready file. Then the cover. I wanted it to feel old, like something you'd find in the back of a library, in the deep Oxford red I remembered from religious books growing up. Give the AI a feeling and a reference point, not just "make a cover."
1. Compile the full public-domain text of [SOURCE TEXT] and format it as a print-ready interior file for a hardcover book: title page, table of contents, clean chapter typography, correct margins for print.
2. Design the cover. Direction: it should feel [FEELING — e.g. "old, like something discovered in the back of a library"]. Use [SPECIFIC COLOR / MATERIAL REFERENCE] as the anchor. Give me 3 variations and we'll iterate from the strongest one.
Step 05
Find a print-on-demand supplier
The agent found Lulu: they print a book when someone orders and ship it straight to the customer. No inventory, and they have an API, so Claude uploaded the files and created the project itself. I never touched a book.
Find a print-on-demand supplier for [PRODUCT TYPE] that (a) prints and ships per-order so we hold zero inventory, (b) ships to the US, UK, Canada, and Australia, and (c) has an API so you can create the product and route orders programmatically.
Compare unit cost and shipping cost per order for the top options, then set up our product with the best one using their API.
Step 06
Price it with real math
My first instinct was $30. The math said that would light money on fire. Here's the actual unit economics Claude ran:
| Print cost per copy | ~$15.00 |
| Shipping per order | ~$7.00 |
| Estimated paid-ads customer acquisition | ~$20.00 |
| Price Claude recommended | $49.95 + $6.95 ship |
| Room left for fulfillment, ads, and profit | ✓ |
Our hard costs are [UNIT COST] to produce and [SHIPPING COST] to ship per order. Estimate a realistic paid-ads customer acquisition cost for this niche, then recommend a price that leaves room for fulfillment, advertising, and profit. Show me the full unit economics at 3 price points, including the one where we'd lose money — I want to see the floor.
Step 07
Build the business on Whop
Day 2. First, turn everything from day 1 into a PRD (a product requirements doc). Then hand that PRD to Claude Code with the Whop CLI installed, and it builds the actual business: products created, payment links live, ready to take money worldwide.
Turn everything we did yesterday into a PRD: the product, the pricing and unit economics, the brand assets, the supplier setup, and what the store needs (product listings, payment links, shipping countries). Write it so a coding agent can execute it without asking me questions.
Using the Whop CLI, create my business on Whop following this PRD. Create the products, set the prices, and give me the payment links when you're done.
[PASTE YOUR PRD]
Whop · Partner link
This is where your business actually lives
Free to start, no subscription, and the API is what lets Claude run the whole thing. Create your account before Step 7 so the CLI has something to build into.
Start your business on Whop →Partner link. I earn a commission if you build on Whop. It's also genuinely what I used in the video.
Step 08
Build a site that converts
Don't design from scratch. I grabbed a physical-product template from the Whop template marketplace and told Claude to swap in my product. Then the important part: research what actually makes cold traffic buy, and turn that into a second PRD.
Research the top e-commerce brands selling to cold traffic. Document the urgency, trust, and social-proof tactics they use on their product pages — guarantees, reviews, shipping badges, scarcity, bundles, whatever is actually on the page.
Then write me a PRD for improving my site's layout using those tactics, specific enough that a coding agent can implement it directly.
Feed that PRD back into Claude Code against your template. A few hours of iterating got me a landing page ready for paid traffic. I turned my finished site into a free template you can grab at thebrettway.com/templates.
Step 09
Research and generate 50 ads
Three prompts, in order: spy, position, produce. This sequence is the difference between generic AI ads and ads built on what's already converting in your niche.
Research my competitors' Meta ad libraries for [NICHE / PRODUCT TYPE]. Make a document listing every tactic they're currently using to convert cold traffic: hooks, angles, formats, offers, and the claims they lean on.
Based on that competitor research, decide how to position our offer so we're clearly differentiated. What angle is underused? What can we credibly claim that they can't?
Make a PRD for 50 ad creatives based on the competitor research and our positioning. Vary the hooks, formats, and angles. Then execute it in Claude Code using the Whop ad generator.
Step 10
Launch the campaign
While the creatives generated, I did the plumbing, all inside Whop: installed the Whop pixel on the site, built a dashboard to track the funnel, and created my Facebook page through Whop with one click, using the same logo and banner from X.
Review the 50 ad creatives with me, then set up a Meta campaign on Whop using the CLI:
- Budget: $50/day
- Targeting: US, UK, Canada, Australia
- Objective: purchases, tracked through the Whop pixel
Set it live from the terminal when I confirm.
Step 11
Fix the funnel
When traffic shows up but sales don't, don't guess. Make the AI diagnose it.
Analyze my funnel. I'm getting [N] visitors from [SOURCE] and zero sales, while paid traffic converts at [RATE]. Walk the full path from click to purchase and tell me the most likely reasons this audience isn't buying. Rank them by impact.
Claude found two problems with my organic traffic: the price was high for a cold audience that didn't trust a sub-100-follower account. So I dropped to $39.99 with free shipping (no ad cost on organic clicks means I can give up margin), added a free audiobook I generated in ElevenLabs, and added upsells: a two-book bundle and an Angel Index. People who buy books buy more books.
Then test your own checkout. On your phone.
Mine was broken and I didn't know. Multiple product bundles made the embedded payment modal take 7+ seconds to load, so buyers clicked and saw a blank page. Days of "why is nobody buying" came down to this.
My embedded checkout takes several seconds to load after someone clicks buy. Preload everything the payment modal needs on first page visit so checkout opens instantly.
Step 12
Crack organic traffic
New accounts have a cold-start problem: my first posts got under 5 views. Instead of waiting months for the algorithm, Claude's move was to pay established pages in the niche to repost our content.
Find 20 X accounts similar to ours in the [NICHE] space with real engagement. Draft a short DM for each asking if they sell reposts. Track who replies and what they charge.
Three replied, one was relevant, $25 per repost, and I negotiated 5 for $100. Reposts got us 10,000+ impressions and 100+ visitors. The first organic sale took until day 7 (after the checkout fix and the better offer), and it was an $80 bundle order. More important: the account now averages 70+ views per post on its own. That's a marketing asset that compounds.
Your turn
Run it for your own product
The whole system is: constraints → AI finds demand → print-on-demand → Whop for the store and ads → diagnose and iterate. It took me 37 hours to the first sale with a product I'd never have picked myself.
Grab the free website template from the video at thebrettway.com/templates, and if you build something with this, tell me in the community. I feature the best builds.
Whop · Partner link
Step one is a free account
Everything in this playbook after Step 6 runs on Whop: checkout, payments, pixel, ad generator, Facebook page, campaigns. Start there.
Start your business on Whop →Partner link. I earn a commission at no cost to you.