Penguins and pasta: What I learned from making an app in 4 weeks with AI
SMRTR summary
A dolphin launches into the air as a toddler squeals with delight. Meanwhile, his dad is quietly shipping code from his phone.
That's the unlikely setup for one developer's attempt to build a full app, backend and all, in just four weeks, while parenting a small child at SeaWorld and the penguin enclosure.
The goal: win a $20,000 prize in the RevenueCat Shipyard competition. The app: a money-saving tool for busy moms, blending batch cooking recipes with upcycling and community features.
Using a terminal-based AI tool called OpenCode, connected via SSH to a virtual machine at home, the developer orchestrated multiple AI agents working in parallel, pushing code, running tests, and deploying builds, all from a phone keyboard, with predictably atrocious spelling.
The verdict on the AI tools? Claude's Opus model impressed. Google's offerings were, in his words, like a high-school project turned in by someone who "didn't really understand the source material."
He submitted. He did not win. But as he noted, at least he doesn't have to pay tax on zero dollars.
SMRTR provides this summary for quick context. The original article belongs to LogRocket.
Read the original article