Writing

Thoughts on AI, engineering, and making things. Less polished than talks, more considered than tweets.

What Makes AI Agents Useful Is Exactly What Makes Them Dangerous

OpenClaw's security failures weren't bugs in one project. They're architectural to the entire category of AI agents. Here's where real solutions are emerging.

Stop Asking What AI Can Do For You

The companies succeeding with AI start with problems, not technology. Here's how to reframe the conversation.

How to Spot Fake Agentic AI

Most products labeled 'agentic' aren't. Here's how to tell the difference.

MCP vs A2A: What's the Difference and When Do You Need Each?

Two protocols, two different problems. A practical guide to understanding when your AI needs tools versus when it needs teammates.