Building Spots.fm in 3 Days: A New AI Startup Paradigm

Justin Jackson recently launched Spots.fm, a platform that lets independent creators monetize YouTube videos, newsletters, or podcasts with sponsorships. He’d been mulling the idea for six years and tried to build it in 2018, but his full-time job got in the way. This time around, he used Claude to design and Fable to code, going from wireframes to a working Ruby on Rails app in just three days—a real efficiency boost for non-technical founders.

Here’s how he did it:

• Design: He sketched rough wireframes by hand and fed them to Claude Design, which generated an interactive prototype. That cut out the usual back-and-forth with a UI designer and gave him visual feedback instantly.
• Product requirements: He used Brian Casel’s plugin to generate a PRD (product requirements document). That step was critical—giving the AI clear business logic constraints kept it from writing messy, unfocused code.
• Coding: He fed the PRD and designs into Fable, which auto-generated an initial version using the Soloterm.com meta-framework. For anyone who understands basic engineering logic, the AI-produced code ran right out of the box.

Why not ship it immediately?

Justin’s goal was never scale; it was “building in public.” He documented the process on social media, driving organic traffic to a waitlist. That Build in Public strategy lets him gather early demand signals at virtually no cost. If search volume is low, signups are weak, or the product feels clunky, he walks away—and the whole experiment costs almost nothing. Only when the data confirms founder-market fit does he consider building a production-grade version alongside his co-founder Jon Buda.

The economics have shifted:

Software used to be expensive because building it required scarce skills. Now AI has flattened the supply side, pushing marginal cost toward zero. When everyone can build software, just having a product isn’t valuable anymore. What matters is distribution capability and how fast you can validate the market. Justin’s playbook is worth copying: don’t sit on a big idea—use AI to throw something usable together quickly, toss it at the market, and see what happens. If most great ideas die from lack of distribution, fast iteration is the only survival strategy. For ordinary founders, this combo—Claude for design, Fable for coding, public validation—compresses the trial-and-error cycle from months down to days, making it the most pragmatic path right now.

Source · Justin Jackson: Read original post →