Read the original at Lenny’s Newsletter ↗lennysnewsletter.com · subscriber post
By Lenny Rachitsky · lennysnewsletter.com · @lennysan on X · YouTube · LinkedIn
Designer Ben Shih describes building LennyRPG, a Pokémon-style pixel game where players meet and quiz Lenny's Podcast guests using transcripts from 300+ episodes. The post walks through his AI-assisted workflow: defining the idea visually, generating a PRD via AI interview, prototyping in Phaser after pivoting from RPG-JS, scaling with CLI tools for quiz and avatar generation, and adding a Supabase leaderboard. It also announces Lenny's release of his newsletter and transcript archive for AI use.
Subscriber post — summary only01Key takeaways
- Spend extra time on the core idea and PRD; these largely determine how smoothly the AI-assisted build goes.
- Let an AI interview you to draft a PRD, then answer its questions, which surfaces gaps in your assumptions.
- Check open source libraries early, but be ready to pivot when a framework fights your core mechanics.
- Ask the AI to log attempts and fixes in a Markdown file so it keeps context as the codebase grows.
- Turn repetitive manual steps, such as enriching data and generating assets, into CLI tools run in bulk.
Summary and takeaways written by PM Atlas; quotes are short excerpts. © the original author.