Real example · Build apps (vibe coding)
Vibe-coded a mini app in an hour, then needed 100 hours to ship it
What they did
Mac Budkowski scoped an idea with ChatGPT and built Cryptosaurus with Claude Opus: a Farcaster mini app that turns a profile picture into a dino-styled image, generated through the Gemini API, which users can mint for $2.
How it went
A working local prototype appeared in the first hour, but the finished product took about 100 hours and roughly 200 image-prompt iterations. He reports 1,000+ downloads, 180+ paid mints and a top-three place in the mini app store in week one, and puts the real speedup at 10 to 100 times.
Worth knowing
Much was manual or went wrong: a launch bug left some payments stuck and he refunded them by script, the cloud setup was overbuilt, and a browser dark-theme bug that three AI models could not fix was solved by a senior engineer in one line. Figures are self-reported.
Try it yourself with Claude
Replace anything in [brackets] with your details.
I want to ship [small app idea] to real users. Before writing code, ask me clarifying questions. Then give a one-page plan: the user flow, what data is stored, the services needed, and the three most likely failure points at launch (payments, sign-in, odd inputs).
Keep the architecture as simple as possible and tell me what you would deliberately not build yet.Read the original
The 100 hour gap between a vibecoded prototype and a working product ↗
kanfa.macbudkowski.com · · Personal post
Attributed to the source linked below; not our own test.
Share Share on X Share on WhatsApp
