Let AI Agents Handle the Refactoring Nobody Wants to Do
Migration tickets sit in backlogs forever because they're tedious, not because they're hard. That's exactly the work an agent should do.
The Ship Room
Build notes from a solo engineer running autonomous agents that ship code while I sleep. Agent architecture, the cost and latency tradeoffs, local-first AI, and where the market is heading, with the real numbers and the dead ends, not just the wins.
Live Last read · last published · next check
Latest posts
The Bottleneck in Enterprise AI Isn't the Model. It's the Context Layer Underneath.
Every team building AI inside a company hits the same wall: the model doesn't know who you are.
The Router Sees Everything: Adding Telemetry to Jano
A model server knows its own last response time. It cannot tell you how deep the queue is, how often the box is thrashing between models, or what every other caller has been getting lately.
The MoE Speedup, Measured: 50 Prompts, Two Local Qwen Models, One Mac Studio
A dense 27B at 8-bit runs my local chat work at 15 tokens a second. Its 35B Mixture-of-Experts sibling runs the same prompts at 70.
Cognitive Debt: The Real Cost of AI-Generated Code
The agent solved the problem and the theory stayed in the prompt. Notes on what gets left behind when the code is clean and the model in your head isn't.
Geofenced Out of the Frontier: Why I Run My Own Models
Two labs fenced their best models behind US compliance in a single week. I build from outside the fence, and I'd already bet on models nobody can revoke.
When the Reader Is a Robot: I Tried to Fix Ads for AI Agents, Then Killed the Idea
For twenty years the web ran on a quiet handshake: free content, search traffic, ads to pay the bill. The handshake is coming apart because the reader is no longer a person. It is an agent.
How Autonomous Coding Agents Are Changing Solo Dev Productivity
The constraint used to be typing speed. Now it's how well you direct, review, and orchestrate. Notes on what one person can ship with the right agent setup.
Almost Sherlocked by Anthropic
On April 8, I shipped an autonomous dev pipeline. The same day, Anthropic announced Claude Managed Agents. Same problem, same time. One is the engine. The other is the self-driving car.
The Three-Body Agent: Orchestrating Agents with GitHub Actions and Claude Code
I moved my entire autonomous agent pipeline from OpenClaw to GitHub Actions and Claude Code CLI. Anthropic's subs decision was the catalyst, but portability and auditability were the real reasons.