I'm building Underpriced AI — an app that tells resellers what their stuff is worth. Snap a photo, get a price from real sold data across eBay, Poshmark, Mercari, and more.
Last 10 days were a blur. Here's the rundown.
The big move: fully automated blog pipeline
I had 51 hand-managed MDX blog files. Painful to maintain, painful to publish.
So I nuked them all and moved everything into the database. Now the entire pipeline is automated:
Discover topic → Generate brief → Claude writes full article → Auto-publish
All triggered from the admin panel. I went from "writing a blog post takes 2 hours" to "click a button, get a published post." Shipped 30 SEO
blog posts and 10 category landing pages in a single push.
The trick was using Claude Sonnet to write from structured briefs (topic, outline, key points, target keywords) rather than open-ended
prompts. Way better output.
Rebuilt the SEO dashboard from scratch
Old version was basically a keyword list. New version:
It's the "marketing command center" I always wanted but never had time to build. Now I do my entire SEO workflow from one admin page.
Mobile app polish
Underpriced runs as a native app via Capacitor. Spent a solid chunk of time on mobile UX:
Not glamorous work, but this is the stuff that makes people actually keep using your app.
Other wins
Lessons from this sprint
What's next
Accuracy is everything for a pricing tool. Next sprint is focused on improving scan quality and getting more user feedback loops in place.
If you flip stuff on eBay, Poshmark, or Mercari — I'd love for you to try it out and tell me what you think. Always looking for honest
feedback from resellers.