Articles
What Happens When Your AI Agent Briefing Lies
The session briefing said 10 alerts. There were 270. Fixing it took four tracks, $741 in compute, and a captain who refused to accept done.
Agents Building UI They Have Never Seen
Seventeen PRs, three editor panels, a navigation redesign, and 40 hours of agent time. The human directing the work has never opened the app in a browser.
Design Specs as Agent Infrastructure
Agents building UI from text descriptions produce divergent implementations. Design specs loaded at startup solved this.
A Design Tool Bake-Off - Figma MCP vs Google Stitch
We tested Figma MCP and Google Stitch head-to-head on the same three UI panels. Sixty API calls versus three.
Cross-Venture Context - Teaching Agents Where They Are
Agents operating across multiple products need spatial awareness. Without it, they target wrong repos and leak secrets across contexts.
When Your Agents Spend 40 Hours on One Auth Bug
Four root causes, 40+ hours of agent time, one MCP auth failure. The final fix was a one-line CLI command.
Tool Registration Is Not Tool Integration
Registering an MCP server in a CLI config means the CLI can discover the tools. It does not mean the tools can access the credentials they need to function.
From Zero to Landing Page in Four Days
A new venture went from repo creation to production-ready Astro site with 46 merged PRs in four days. The infrastructure did most of the work.
Taking Product Development Offline with Local LLMs
We set up four specialized local models on a fanless laptop for offline product work. Here is what we built, what works, and what we are measuring.