Developer Builds Blog Feature on Phone Using Claude Code

A Bird Photography Feature, Built Entirely on Mobile

Developer and AI tools advocate Simon Willison has demonstrated yet another practical use case for AI-powered coding assistants — building a fully functional blog feature entirely from his phone using Claude Code for web.

The feature, called 'Sightings,' pulls Willison's wildlife photography from iNaturalist and syndicates it directly to his personal blog. The project was inspired by a new Canon R6 Mark II camera and a growing passion for bird photography.

From Prototype to Production in a Day

Willison shared that the feature went from a successful prototype to a shipped pull request in remarkably short order. He built it as an extension of his existing 'beats system' — a framework he uses for syndicating external content into his blog.

The sightings now appear across multiple surfaces on his site, including the homepage and date archive pages, treated with the same first-class status as his other syndicated content types.

What makes this notable isn't the feature itself — it's the workflow. Willison used Claude Code's web-based interface directly on his phone to write and submit the code, demonstrating how AI coding assistants are making development possible in contexts that would have been impractical just a year ago.

Claude Code Goes Mobile

Anthropic's Claude Code has been gaining traction among developers as a terminal-based coding agent. The web version extends that capability to browsers, and as Willison's example shows, even mobile browsers become viable development environments when paired with a capable AI assistant.

Willison publicly shared both the pull request and the prompt he used, offering transparency into how he directed Claude Code to produce the working feature. This kind of open documentation has become a hallmark of his approach to AI tooling — showing not just what's possible, but exactly how it's done.

Why This Matters

The broader significance lies in the shifting boundaries of what counts as a 'development environment.' Mobile-first coding has long been dismissed as impractical, but AI coding assistants are changing that calculus. When the AI handles the bulk of syntax, file management, and boilerplate, the developer's role shifts toward architecture and intent — tasks that don't necessarily require a full desktop setup.

Willison's experiment also highlights a growing trend of developers using AI tools for personal projects and side features, treating them as low-friction ways to extend existing systems without dedicating large blocks of focused coding time.

Looking Ahead

As AI coding assistants like Claude Code, GitHub Copilot, and Cursor continue to mature, expect more developers to adopt mobile and on-the-go workflows for lightweight feature development. The gap between 'having an idea' and 'shipping code' continues to shrink — and examples like Willison's sightings feature show that gap can now be closed from a phone screen while presumably watching birds.