How I Use Claude AI Agents to Build My Entire POD Business in 2026
About this video
RJ Martinez demonstrates building his print-on-demand t-shirt business around a Claude AI agent in the Claude desktop ‘co-work’ app, which he says he runs on the $100-a-month Max plan. He dictates a prompt with a voice tool (Whisper) asking the agent to scrape Reddit, X and YouTube for newer, low-competition t-shirt design concepts that will sell, and to maintain them in an updatable spreadsheet; the agent first asks clarifying questions about audience (he picks hobby/identity, pop culture and memes), sales channels (multiple platforms) and refresh cadence (he chooses twice a week), with trademark and copyright checking built in. The agent runs sub-agents, builds a color-coded ’t-shirt design idea tracker’ — he says it sourced 28 research ideas from Reddit, YouTube, TikTok, X, Pinterest and Instagram, with source links, SEO keywords, price ranges and trademark flags — and sets a scheduled refresh (Monday and Wednesday at 9 a.m. in his setup). He then iteratively prompts it to add actual social-media source-post links, four to five concrete design ideas per concept (text-based, text-plus-graphic, or graphic), and a ready-to-paste image-generator prompt for each design, working around Reddit/YouTube blocking and a daily usage limit along the way. He pastes a prompt into Ideogram, generates an AI-humor design (‘I asked AI to draw a horse and I got this’), downloads it on a transparent background and upscales it. Finally he builds a store with EverBee: AI generates the store (named ‘AI LOL Gear’), he connects Printify through the app marketplace, uploads the design onto a Comfort Colors t-shirt, reviews EverBee’s listing optimization and pricing (the example lists at $24 against production costs he puts at roughly $12–17 plus $4–5 shipping), publishes through Printify, syncs and publishes the product to the live store, and previews the product page. He stresses the same agent workflow applies to any business, not just t-shirts, and that designs and trademark checks still need human fine-tuning.
Key takeaways
- Use a Claude desktop agent with a dictated, detailed prompt to scrape Reddit, X, YouTube and other platforms for low-competition design niches and keep results in a structured spreadsheet.
- Build trademark/copyright checking and a scheduled refresh (the creator chose twice weekly) into the agent workflow from the start.
- Extend the spreadsheet iteratively: source-post links, 4–5 design concepts per niche, and a ready-to-use image prompt for each design.
- Generate the design in Ideogram with a transparent background, upscale it, and push it to a product via Printify connected to an EverBee store.
- The creator frames the agent as a reusable employee-like workflow for any business, but says human review of designs and trademark flags is still required.
Episode details: published March 31, 2026 · 26:40 long. Summary written by Self Growth Videos from the episode itself. Business results and any income figures mentioned are RJ Martinez’s own claims, shared for education — they are not typical or guaranteed.
← Previous: Amazon Hot New Releases Research for Print On Demand 2026 · Next: 5 Low Competition Print On Demand Niches Selling RIGHT NOW →
All RJ Martinez videos · RJ Martinez profile · RJ Martinez on YouTube
You Might Also Enjoy

I Ran Etsy's New AI Test on My Own Shop. One Listing Failed.
Explore →
This Etsy Shop Made 48,900 Sales Without Designing the Art (Steal This Store #5)
Explore →
I Hired an AI Employee for My Etsy Shop - Here's Where It Messed Up
Explore →
I Ordered 8 PrintKK Products - 3 of Them Most Sellers Don't Know Exist
Explore →
Someone Might Be Selling Your Design Right Now (Here's the 5-Minute Check)
Explore →