Start here
New to Antinode Labs? These posts give the clearest picture of how we think about AI agents and engineering. Read them in order.
-
The Five Levels of AI Coding Agents: A Practical Rating - From Autocomplete to “Just Ship It”
A personal take on AI coding tools—from Copilot’s autocomplete to Claude Code’s autonomous engineering—and the five levels that actually help you decide what to trust
-
Your AI Isn't Broken. Your SDLC Is.
We shoved AI into the coding phase and called it progress. Meanwhile, design, testing, and operations are still running on 1990s logic. Five hard truths about why you're not shipping faster and what to do about it.
-
The Half-Built House: Why Software Teams Start But Never Finish (And How to Stop)
A practical guide to escaping over-engineering, restoring team momentum, and delivering value—one small win at a time.
-
The ADLC: What I Got Wrong About Agents, Entropy, and the Code I Couldn't Read
Six months after admitting better prompts weren't the problem, I discovered a deeper one. AI agents generate code faster than we can understand it. Here's the framework I use now.
-
Why Is More Context Making Your AI Agent Worse?
Context is what you reach for when you can't articulate what matters. The alternative is harder, and it's the only thing that works.
Looking for more? Browse the archive or all tags.