AI-agent-run Print-on-demand Book Business Built on Public-domain Texts
Originally published Aug 27, 2026 — Brett Malinowski
Brett Malinowski · I Gave AI 7 Days to Sell Ethiopian Bibles ↗
Use AI agents (Claude) to find an out-of-copyright text with existing online demand (the presenter's example: the Book of Enoch/Ethiopian Bible), turn it into a hardcover physical book, print and ship it per-order through a print-on-demand supplier (Lulu.com) so there's no inventory to buy upfront, and run the business (website, ad creatives, campaign setup, funnel fixes) largely through AI coding/agent tools.
Malinowski gave an AI agent a $500 budget and three rules (AI makes every decision, AI generates all content, $500 total budget), and had it find the product, build the site, generate ad creatives, and run Meta ads. The first sale came within 37 hours; after a full day, gross revenue was $232.15 against $58.55 in ad spend and $119 in printing/shipping, for $54.60 profit. Over a full 7-day test the business generated $745.25 in total revenue, with organic distribution (a themed X/Twitter account, plus paid reposts from established niche accounts) eventually producing its first non-ad sale as well.
Someone comfortable directing AI tools (or willing to learn Claude Code/agent workflows) to do the legwork of research, copywriting, and ad creative generation, who wants to test a low-inventory-risk physical product business without doing every step by hand.
The core mechanic is genuinely replicable and well-documented step by step (public domain text, print-on-demand fulfillment, AI-assisted marketing) and doesn't require deep technical skill, just willingness to direct AI agents through the process. That said, this is one creator's single 7-day experiment with a $500 test budget, not a proven scalable business, and the presenter says outright he's giving the whole built business away in a giveaway rather than continuing to run it himself, a signal this was framed as a demonstration/content project first and a business second.