Mac App in 11 Days Without Code Using Claude
Alex da Cruz
Alex da Cruz is a full-stack developer based in São Paulo, Brazil. He works with React, TypeScript and automation, and uses AI daily to solve real problems in code and operations — not as a demo. He has run an e-commerce operation end to end, and now builds and maintains the automation pipeline behind this blog. He writes about what he actually tests.
Building software without writing code demands active supervision rather than magic prompts. According to ZDNet, developer David Gewirtz used Claude Code to build and launch a paid Mac application in just 11 days, taking four days of active interaction split across 502 individual directives.
How many steps does vibe coding actually take?
The process requires treating the artificial intelligence like a human partner rather than an oracle. Instead of relying on a single command, the developer guided Claude through strategic decisions, debugging slow window scrolling caused by redundant image compression, and implementing security measures to prevent users from easily copying raw assets from the package contents.
What are the practical hurdles of AI-driven app stores?
Beyond generating functional code, the project faced strict compliance and ownership checks. Claude actively questioned the provenance of vintage icons to ensure the app would survive Apple's review, leading to the removal of outdated or potentially copyrighted assets like a Caterpillar logo and gun icons before submission.
Sources
Frequently asked questions
- How long did it take to build the Mac app?
- The entire process from concept to App Store submission took 11 days, with four days of active development involving Claude Code.
- How many prompts were used to create the application?
- The developer issued 502 individual directives to Claude Code while managing other daily professional and personal priorities.
- How were app assets protected from piracy?
- Claude helped encrypt and package the app with a single scrambled file, decrypting images dynamically when requested to prevent users from extracting raw files.
Comments
0 comments
Be the first to comment.
Continue Lendo
Mercor Study: AI Outpaces Accountants in Speed
A new Mercor study shows AI models now beat licensed accountants in speed and accuracy on structured tasks, though supervision remains mandatory.
Meta Muse AI Agent: Workflows, Pricing, and Security Controls
Meta enters autonomous AI execution with Muse. Here is how its API connectors, pricing tiers, and virtual security controls function.
Reka Rho-1: Omnimodal AI Processes Text, Video, and Robots in One Network
Reka AI introduces Rho-1, a 19-billion-parameter omnimodal model combining text, video, and robot control in a single context window.