I challenged VS Code, Claude Code, and Google Antigravity to build an app in an hour, but only one was fit for the Play Store
An experiment published on 2026-11-16 challenged three AI coding tools, VS Code with GitHub Copilot, Claude Code with Opus 5.5, and Google Antigravity with Gemini 3.8, to develop a functional Android application within a 60-minute deadline. The test, which involved building a subscription management app, found Google Antigravity’s Gemini 3.8 to be the most effective in balancing user interface and experience design for a Play Store-ready product.
An experiment assessed the capabilities of three AI coding tools in developing a functional Android application. The challenge required VS Code, Claude Code, and Google Antigravity to create "SubTrack," a subscription management app designed to track recurring payments, complete with a dashboard, spending insights, renewal dates, and subscription management features.
Each tool was given identical, detailed prompts and a strict 60-minute deadline, with no allowance for additional instructions or manual code adjustments. GitHub Copilot was used for VS Code, Opus 5.5 for Claude Code, and Gemini 3.8 for Google Antigravity. The resulting Android Package Kits (APKs) were then installed and evaluated on a Pixel 8 device to assess their readiness for the Google Play Store.
VS Code, leveraging GitHub Copilot, produced a functional app with a clean interface and a dark theme. While its core subscription management features worked, the experiment noted "questionable UX decisions," such as an overlapping floating `+` icon and an "Add subscription" button, along with a basic New Subscription menu. These design oversights suggested the app was not yet suitable for the Google Play Store without further refinement.
Claude Code, powered by Opus 5.5, delivered the most visually impressive user interface, featuring a modern design, attractive cards, and a dashboard that felt premium. The core features also functioned as expected. However, the experiment identified less attention to detail in user experience, including a `+` icon placed inconveniently in the top-right corner on tall smartphones, repetitive "Add" buttons, and generic placeholder icons instead of recognizable service logos.
Google Antigravity, utilizing Gemini 3.8, emerged as the top performer in the experiment. Its generated app struck a balance between a polished user interface and thoughtful user experience design. The interface was described as polished and deliberately designed, without the awkward button placements or alignment issues observed in the apps from Copilot and Claude Code. This outcome positioned its generated app as the closest to being ready for public release on the Google Play Store under the strict test conditions.
Share this article
Related reading
6 stories
Meet Boris Cherny, former Meta engineer behind Anthropic's Claude Code
Meet Boris Cherny, the former Meta engineer behind Anthropic's Claude Code, one of the AI tools challenged here.

NETS Explores a Way for AI Agents to Pay at Singapore Shops

OCBC Aims to Train 20,000 Small Business Owners and Staff for Free

How a little-known Beijing AI safety group became an adviser to China’s tech giants

Nium to Handle Real-Time Overseas Transfers for Woori Bank

