I Asked AI To Make Me As Much Money As Possible In 10 Days (RAW RESULTS)

Summary

The creator tested whether an AI agent could run an Etsy business for 10 days, with no ads or audience promotion. The AI researched, created, and listed digital artwork; the creator checked in once daily and adjusted the strategy. The experiment produced two orders and $4.55 in gross revenue, but the creator said it was too early to know whether the shop could cover its costs.

Business model and operating setup

• Product: Digital PNG artwork for customers to download and use, including seasonal designs and themed bundles.
• Sales channel: Etsy, chosen because it already has buyers searching for products and supports digital downloads.
• Brand: Little Wonder Studio AU, presented as a cozy Australian art shop.
• Division of work: GPT-6 Astra, operated through Codex, handled research, artwork, listing creation, and store updates. The creator supplied direction, reviewed performance, and changed the workflow.
• Experiment constraints: No advertising or creator-audience promotion; the goal was to attract organic customers.

Playbooks and tactics tested

• Reduce production cost per listing: The first listing used about 4% of the weekly Codex usage allowance. After iterative workflow optimization, the creator reported reducing this to 0.33% per listing, raising theoretical capacity from roughly 25 to more than 300 listings per week.
• Use keyword research for customer acquisition: The workflow incorporated eRank to compare Etsy search demand with competition. The creator cited Christmas PNGs as highly competitive—about 11,000 monthly searches versus 842,000 listings—and Mahjong PNGs as a more promising opportunity, with roughly 1,000 search results against 642 listings.
• Increase listing volume through automation: The initial target was about three listings per hour, with a later goal of 100 listings in 24 hours. The creator believed AI made this volume possible for a solo operator, though volume alone did not guarantee discovery or sales.
• Test bundles and discounts: The AI was directed to bundle related designs and experiment with discounts from 10% to 90%, using common Etsy merchandising tactics. The video does not establish which discount level worked best.
• Iterate from shop signals: The creator monitored views, visits, customer messages, orders, and abandoned baskets, then adjusted keywords, bundles, discounts, and listings.

Results and metrics

• Listing output: 191 listings by the end of the experiment.
• Final traffic: 48 visits.
• Orders: 2.
• Gross revenue: $4.55—one order for $1.16 and another for $3.82, before Etsy fees and AI operating costs.
• Abandoned baskets: 8. The creator identified abandoned-cart discounts as a possible follow-up tactic but did not implement them in time.
• Early signals: After the first day, the shop had 14 views and 7 visits, with no orders. Later, it had 37 views and 18 visits and received a customer question about whether a PSD file included layers. The shop did not offer layered files, revealing a potential mismatch between the customer’s expectations and the product.
• The first sale came from Halloween piglet artwork; the second was a Halloween sheep bundle.

Business lessons and caveats

• Listing volume is not the same as traction. The creator initially treated the number of listings as the main scoreboard, but the experiment showed the need to give customers a clear reason to buy and to convert shop traffic.
• Automation still needed human direction. The AI produced research, art, and listings, while the creator reviewed results and changed the strategy.
• Measure unit economics before calling it a business. The $4.55 result was gross revenue; fees and the cost of running the AI were not deducted.
• Protect intellectual property. The creator’s previous Etsy shop generated more than $10,000 over six months and was making about $600–$1,000 per month, but Etsy shut it down because it sold Harry Potter stickers without commercial rights. The creator cited this as a reason to avoid repeating the mistake.
• The creator planned to continue working on the shop, but the 10-day test did not demonstrate a profitable or sustainable business.

Presenters and sources

• Presenter: The unnamed creator of the Dubibubi channel.
• Tools and platforms discussed: GPT-6 Astra, Codex, Etsy, and eRank.
• Other source cited: The creator’s prior experience running an Etsy shop.