No-one should waste time on bad films. With limited viewing time, spending 40 minutes browsing streaming archives filled with mediocre content is inefficient. I was already checking reputable sources before committing to anything. The solution was to automate that behavior and integrate it with streaming availability.

The product doesn’t optimize for engagement, like streaming platforms do. It focuses on quality based on critical consensus: awards, festival selections, critical acclaim.

First, I worked with Claude to write a comprehensive brief outlining the actual film discovery behavior I wanted to support and narrowed the scope down to a manageable MVP. I fed that to Claude Code and it quickly generated the initial structure.

Although there was a lot of room for improvement, it was amazing to have an initial working version of the product in a few minutes. As a designer, I’m used to finding flaws in a design by drawing wireframes and tapping on rudimentary prototypes. Claude Code took this to another level – it built something I could proudly put in front of stakeholders for a quick feedback round.

While I worked on the visual design in Figma, Claude Code powered through my requests for integrations and improvements.

I built custom agents to verify security, UX, copywriting, and accessibility. Using Figma MCP, Claude could read my frames directly. With Playwright, it could take screenshots, implement code, then validate the browser output and iterate until the implementation matched the design.

Claude Code proved significantly more thorough than Cursor for this project. It identified and resolved usability gaps in my initial designs, often delivering improved versions. It functioned as a design partner, proposing solutions when I hit constraints. It also iterates on code until it works correctly rather than marking tasks complete with unresolved bugs.

I had limited prior experience with production deployments and PR workflows. This project required deploying a full application to production and addressing security, infrastructure costs, and naming conventions. The experience was valuable and I’m continuing to expand my technical capabilities through iteration.

The platform has operational costs, so monetization is the next priority. I’m currently evaluating several approaches for implementation.

The recommendations have been consistently accurate in personal use. Several friends are using the platform regularly. There’s a substantial list of improvements planned, but the core product validates the initial hypothesis.