I Gave AI 7 Days to Sell Ethiopian Bibles

0:00I gave AI one goal, make money selling a physical product as fast as possible. I have 10 years of experience running online businesses and have made over $5 million. So, I want to find out if AI [music] is actually smart enough to run a profitable e-commerce store on its own or whether us humans still have an edge.

0:18I'm going to let the AI do everything. It's going to find the product, build the website, run the marketing, and fulfill the orders. Now, [music] I have three rules for the challenge. First, AI makes every decision. Second, any content I create has to be generated by AI. And third, I have a total budget of $500, which I put on this virtual card and gave to the AI agent. To do this, I'm going to use the Claude desktop app and WOP's [music] CLI. Claude lets me use multiple agents for different roles in the business, and lets me build my website, accept payments, [music] and run ads. To start, I gave Claude a

0:50pretty simple objective. Find me a physical product that we can launch quickly with three [music] characteristics. One, there needs to already be demand. Two, I don't want to buy inventory upfront. And three, there needs to be some unique way that we can compete. I do not want the AI inventing some new garlic peeler and then us spending all of our $500 trying to prove that people want it. I want something that people already care about. [music] And Claude landed on something that I never would have expected. The Book of Enoch. Apparently, there is a massive amount of content around the Ethiopian Bible, lost ancient texts, and

1:21specifically the Book of Enoch that are currently getting millions of views online right now. So the demand is already there and there are versions of the book of Enoch that are in public domain which means it's legal for anyone to sell. But this also means that [music] anyone can read it online for free. So the value that we're creating

1:37isn't access to some secret information. It's [music] the actual physical product. Anyone can pull up the text on their phone. But there is something different about having an old text [music] in physical form sitting on your bookshelf. So this gives us a pretty simple business model. Take a text people are already interested in. Turn it into a beautiful hardcover physical book. print each copy on demand so we don't need inventory and use all of the existing attention around the book at Enoch to sell it. So Claude had found our product, but before we start making it, I wanted Claude to start working on distribution from day one. Based on our

2:08budget, Claude told me that organic content would be best and that we should start by posting on X. We came up with the brand name Enox Archive and generated a logo and a banner. Then I asked Claude to research what kind of posts were already getting attention and to make a posting strategy doc. Claude created three scheduled tasks where every day the agent would post five times, reply to 10 popular posts on the home feed, and make two quote tweets. If we build this theme page correctly, we should get a steady flow of organic

2:34traffic without spending any money. We'll check back on this account in a few days. Now, we need an actual book to sell. Claude helped compile the public domain text and format it into something that we could print. Then, we needed a cover. I wanted the book to feel old, almost like something that you would discover in the back of a library. And I wanted to use this dark Oxford red because growing up a lot of the religious texts that I saw had these dark red covers. Claude came back with a design and honestly it actually looked good. And after a few iterations we

3:03landed on a cover that I was happy with. Now we need a supplier. So I created a new task and the agent found lulu.com. They can print a book when someone orders it and ship it directly to the customer. Which means instead of buying a 100 books upfront, we only pay to manufacture the book after someone has already paid us. And even better, they have an API. Claude uploaded the files, created the project, and we now had a supplier that can manufacture and ship the product without me having to touch a single book. It costs roughly $15 to print one copy. Shipping is around another $7. So, our expected hard costs

3:37are $22 per order. Claude estimated that acquiring a customer through paid ads might cost around $20. So, if we sold the book for $30, we would basically be lighting money on fire. Claude recommended we price the book at $49.95 plus $6.95 for shipping. At that price, if the $20 acquisition estimate is right, we would have enough room for advertising, fulfillment, and hopefully some profit. That was a long first day, so I got some rest. On day two, it was time to turn the product into a business. I told Claude Co-work to turn everything we did yesterday into a PRD, aka a product requirement doc. Then I

4:12went over to Claude Code. I installed the CLI and I asked Claude to create my business following the PRD that built the product and gave me payment links so I could accept payments from customers all over the world. Now [music] I just needed a website. So I selected a physical product template from the template marketplace and told Claude to update it with my product details. After a few hours of iterating, I had a landing page that was ready for customers. Now it was time to run paid ads. I had co-work research all of my competitors meta ad libraries and make [music] a document listing all the

4:42tactics, offers, and positioning they are currently using to convert cold traffic. Claude then decided how to position our offer so we were differentiated. [music] Finally, I asked Co-work to make a PRD for 50 ad creatives based on everything we just learned. Now, to actually make those ad [music] creatives, I went back into Claude code, pasted in the PRD, and asked Claude to use the ad generator. While that was happening, I installed the pixel on my website, created a metrics dashboard [music] to track my funnel, and created a Facebook page through I just clicked this button, used the same logo and banner as the Twitter account, and hit create

5:17Facebook page. [music] I reviewed the 50 ads, and they looked amazing. I then told Claude to set up a meta ad campaign on using the CLI. We set the budget at $50 per day, targeted the four major English-speaking countries, and set it live all through the terminal. Then I went home because my brain was [music] fried. After eating dinner with my wife and spending some time with my daughter, I decided to check the ad account before going to bed. And my mind was [music] blown. The ads had been running for under 3 hours. We'd only spent $4.68 and it was already showing that we got a

5:48sale [music] for $56. I really couldn't believe my eyes. So, I went into the account and there it was, $56 [music] sitting in the account. I have started over a dozen businesses in my life and I have never [music] once been profitable on my first day of running ads. So, this challenge was to see how fast AI could go from idea to first sale. And we had just done it in under 37 [music] hours. But this could have been one lucky sale. So, I decided to let the ads keep running all day for day three. And by the end of the day, we'd actually gotten three more sales. So, [music]

6:22after a full 24 hours of running ads, our gross revenue was sitting at $232.15, we spent $5855 [music] on ads and another $119 on actually printing and shipping the books, leaving us with $5460 [music] in profit. That's how much I made for a 4-hour shift working at a pizza shop in [music] high school. And AI just did that in 24 hours. I'll let the ads keep running at $50 per day and we'll check in again on day 7. Now that ads were working, it was time to check in on X and [music] see if we can get our first organic sale. After 3 days, the account

6:58was getting basically no traction. Most posts have fewer than five views. So Claude had another idea. [music] New accounts have a cold start problem. So instead of waiting months for the algorithm to pick up our posts naturally, why not pay an established [music] page in the niche to repost our content? I had Claude find 20 accounts similar to ours and DM' them asking if they do paid repost. [music] Three replied, but only one was actually relevant. They wanted 25 per post. I negotiated five for $100. The first repost got more than 500 views and sent around [music] 10 people to the site. We got no purchases, but now at least we

7:35know that X [music] can send traffic. The next post crossed over a,000 views, and one of our replies had actually got over 8,000 views. So, we got over 10,000 impressions and over a 100 organic visitors, but still zero sales. And that didn't make sense because roughly 100 visitors from ads resulted in three sales just the day before. So, I was very confused why none of these people were buying. So, on day five, I asked Claude to analyze the funnel. It came back with two problems. The price is high for a new audience and people don't trust the product is real, which makes sense. We had a brand new account with

8:12under a 100 followers selling a $50 book. So, I lowered the price to $39.99 and included free shipping. Since I'm not paying for organic clicks, I can afford to give up some margin here. Then, Claude suggested we add free bonuses to increase the perceived value. So, I made an audiobook version in 11 Labs and included it for free. Then, we added some upsells, a twobook bundle, an angel index, and another lost book. The logic is simple. People who buy books buy more books. By day six, I was feeling confident. We had drastically improved the offer. So, I got two posts reposted that day. One got over 1,000

8:49views and another got 5,000 views. I was super hands-on with replies and we actually got over 20,000 views from replies alone. We gained 32 followers and we got another 100 website visitors, but still no sales. I was genuinely about to lose my mind. Like, I was so frustrated. There is no way I don't have

9:08an organic sale with this much traffic. So, I decided to test the funnel and it turns out our checkout was broken. [music] Since we created multiple different product bundles, our embedded checkout modal was taking over 7 seconds [music] to load. So, basically, people clicked by, but then it just looked like an empty page. So, I told Claude to just preload everything when they first [music] visit. And that fixed it. So, after banging my head against the wall all day and burning two reposts, I went to bed kind of angry. [music] Day seven came around and this was my last hope. I had one repost left, so I used it on a

9:41tweet and made sure to include the website link directly in the post. The post started slow and I couldn't look. But 2 hours later, I opened my laptop out and [music] there it was, our very first organic sale. And this customer didn't just buy one book, they bought the two book bundle and the Angel Index order bump, an $80 [music] order. Now, we still weren't profitable on X, but the account now had over 100 followers, and my [music] organic posts were getting around 70 views on average without any reposts now. So, we had successfully built a proven marketing asset [music] that will now compound over time, which is a win in and of

10:17itself. And after the full 7 days, we generated $74525 in revenue. [music] So, the challenge was to see how fast AI could go from idea to first sale. But now I want to give this business away to one of you to scale. You'll get the website, the product, the ads, even the X account. There is a link in the description if you want to enter the giveaway. And even if you don't win, I turned the website I made in this video into a free template that any of [music] you can use. That is the second link in the description below. As I reflect over my last 10

10:46years, I've pretty much started [music] every online business that there is. So, I want to make this a new series [music] on the channel where every few weeks I start a new business and see how fast AI can start making money with [music] it in hopes that it helps you make your first dollar online. So, let me know in the comment section what business you

11:04think we should do next.