Wrangling a Browser Agent to Chart a Benchmark Leaderboard
Claude uses a browser agent to open a benchmark leaderboard, then iterates through several failed attempts to inject percentage labels onto a live chart.
Related
Inside an AI Agent Swarm That Built a Browser Engine
A summary of an interview on an experimental browser rendering engine built almost entirely by thousands of concurrent AI coding agents over a single week.
Document Intelligence Platform
A marketing landing page pitching an AI agent system for document processing, walking through case studies for financial services and legal workflows.
A client-side MP3 silence cutter, built waveform-first
A four-round build of a browser-only MP3 silence trimmer, adding zoomable dual waveforms, click-to-play scrubbing, and neatly categorized CSS along the way.
A deliberately over-specified product management crisis
Asked for an example a thousand times more specific than usual, it invents a latency crisis with six stakeholders, exact penalties and a 72-hour deadline.