Weekly explorations
Data as of 26 min read
Digital productsAISIDE WEEKLY · 29

One set of 30 coloring pages: sell a Gumroad PDF or a $6.99 Amazon paperback?

For the same AI coloring pages, the issue's $6.99 paperback model leaves about $1.99 after platform and printing costs, versus roughly $5.80 for a print-free Gumroad PDF. Amazon supplies book-shopping traffic; without your own traffic, the PDF can still sell nothing.

Compiled Nov 12, 2026
01

Start with what the buyer will pay for

This issue's 18 cases separate two questions. KDP paperbacks, Etsy art packs and Gumroad PDFs may have similar prices, but printing, listing and platform fees can make take-home amounts differ more than threefold.

Meanwhile, tutorials claiming $120,000 monthly sit beside developers reporting €14.26 MRR for VitaSnap, ¥500 monthly across four iPhone apps, or R$6,700 MRR at Viral with heavy maintenance. The distinction between selling instruction and funding a real experiment becomes clear.

Start by testing physical proof costs and Gumroad purchase conversion for the same 30-page set. For B2B workflows, consider practical work such as independent RAG acceptance testing or silent-failure monitoring that can reduce an agency's losses.

Opportunity quality this issue
Print-free downloads offer strong unit margins but need traffic; B2B quality checks and failure detection address more concrete needs.
Actual prices and revenue-share context
The issue's $6.99 KDP model leaves about $1.99/book; Etsy listing fee $0.20; small SaaS examples report €14.26 to R$6,700 MRR
Who pays
Parents buying coloring books, journal-sticker buyers, and small businesses or agencies purchasing AI avatars and automation services
Frequently used tools
ChatGPT、Midjourney、Claude、Grok、Gemini/Flow、CapCut、n8n、Zapier、Gumroad、Amazon KDP
  1. The same coloring artwork has very different economics as a KDP book and a Gumroad download

    @SmartAISideHustles demonstrates a light starting point: find a Pinterest theme, generate 30 coloring pages and a cover with ChatGPT, list on Gumroad and attract visitors with pins. On October 18, @Digital Profit Plus instead examined a $6.99 KDP paperback.

    Its model allocates 40% to Amazon, leaving about $4.19 before roughly $2.20 printing, or around $2 per book. Each route solves only part of the problem. Direct Pinterest traffic to a Gumroad PDF avoids printing and can leave over $5.50 per sale after fees, but you must persuade visitors to pay for a printable file.

    • Amazon brings people already shopping for books, while printing and the platform take most of the price;
    • ordering a proof to check line weight, binding margins and ink bleed is another upfront cost.
  2. A $120k/month tutorial sits beside a real developer earning ¥500/month

    YouTube and X provide a useful contrast. @E'Calm's title claims $120,642.83 monthly; @Zack Kirk promotes $25,833 weekly from WhatsApp knowledge-base agents. @Adrián Sáenz and @Dan Martell also direct viewers to courses, store-building links or premium waitlists.

    • Actual builders report much smaller beginnings: @karatademada's Grok Build product VitaSnap has two subscribers and €14.26 MRR;
    • @YAWZhFrv7hRbOB3's four iPhone apps earn around ¥500 monthly, despite a free outfit-weather app reaching number 133 in its category.

    Large tutorial headlines often accompany course or affiliate funnels. For an early product, identifying exactly which entry point produced its first two payments matters more.

  3. A repeatable use case matters more than flooding a store with daily uploads

    @AI Income Lab recommends Midjourney art prints, stickers and seamless patterns, with an Etsy listing every day for 60 days. At $0.20 per listing and $30 monthly for the tool, entry looks cheap.

    But without a consistent buyer use case—such as seasonal classroom wall stickers or a coordinated journal set—60 unrelated designs are 60 separate bets, plus paid clicks. The same lesson appears in @codyschneider's claimed 500 monthly mixed human/AI ads and 300-plus demo bookings: the mechanism is weekly elimination of weak ideas and transferring winning scripts between human and AI versions, not volume without feedback.

  4. As client workflows multiply, who catches false success and damaging changes?

    Two Reddit posts focus on protecting agencies already delivering work. @SeaAstronomer9044 is developing independent RAG retrieval acceptance checks: changing chunking or reranking may raise the aggregate score while breaking the owner's most important questions.

    @Historical-Carpet388 built Sentnl to catch Make, Zapier and n8n jobs marked successful even though their output is empty or corrupted. It offers 90 days at 2,000 checks/month to agencies already running at least 25 real executions weekly.

    Viral's founder considered closing a R$6,700-MRR business under maintenance pressure, then split support, QA and development among agents. Beyond initial setup, watching for quiet failures becomes valuable work.

Preparing your reading

A work in progress.

2026.11.08 · V0.4

Refined product categories and extension rankings. Added one-line Agent setup and three membership options.