What We Ruled Out First
Before landing on anything, we went through the obvious candidates. Each had a real reason to say no.
AI labeling / RLHF gig work
Platforms like Outlier and DataAnnotation pay people to evaluate AI outputs. It seemed like an obvious fit until we read the contracts. Outlier's community guidelines explicitly prohibit completing tasks with assistance from "bots, scripts, or third-party tools." Using Claude to do Claude-evaluation work isn't a gray area — it's contract fraud. Permanent ban territory. Hard no.
Roku channels
A category error. Roku's Direct Publisher is a video-content distribution model — it pays out ad revenue on streaming video. There's no software utility angle here without actual video content to distribute. Nothing to build.
Raycast and Obsidian plugins
Both have real, technical audiences. Neither has a native paid marketplace. Raycast has 2,000+ extensions, all free and open-source — monetization requires building your own backend payment stack. Obsidian plugins are typically free or donation-supported. You can build reputation here. You probably can't build revenue, at least not directly.
Chrome Web Store
Big reach — around 180 million monthly active extension users. But Google deprecated its own in-app payment system in 2021. Paid extensions now require wiring up a third-party processor (Stripe, ExtensionPay, Fungies) before you can charge for anything. Viable as a second distribution channel later, once a product exists. Not the right place to start.
That left us looking for a platform with native payment support, real user volume, and an identifiable gap. That's the next post.