Posts

If you follow conversations about AI engineering, you have probably seen the word “harness” everywhere lately. Mitchell Hashimoto named the discipline, LangChain dissected its anatomy, HumanLayer called our failures a skill issue, Anthropic published how it designs harnesses for long-running work, and Addy Osmani pulled the threads together. I think this is genuinely good news. Harness engineering is becoming a real discipline, and an important one. As more people study it, document it, and share what they learn, the practice will continue to improve. ...

I have always been an early adopter of new technologies. As Barney Stinson once said, “new is always better.” By the time ChatGPT came out, I was already using GitHub Copilot beta. Before that, I had used Tabnine. I had also been following OpenAI for years, first through their Dota 2 reinforcement learning project, where their agents beat some of the best players in the world. Later, I used their Davinci models in the OpenAI Playground to help with essays and document completion. So when ChatGPT arrived, I did not need much convincing. I jumped in immediately. ...

Before I write about how I am mitigating the issues I have with AI, I have to talk about the issues themselves. There are many great writings out there, that explains the shortcomings of AI agents, such as Addy Osmani’s 70% Problem, Armin Ronacher’s Agent Psychosis, and these tweets by Mitchell Hashimoto: Companies with AI Psychosis and Fake Optimization. So, in the interest of not repeating, and with a confession that I don’t have concrete shareable examples for every single item I have seen, but trusting PG’s point in How to Know, I am going to share what I felt here. ...

In my opinion, workspace hygiene is the single biggest factor that determines what number actually sits in front of the “x” when people talk about Nx engineering with AI agents. Prompting skill and model choice matter, but they are secondary. A clean, well-structured workspace with strong guardrails, clear contracts, reproducible environments, and reliable memory dramatically reduces how often you have to step in and correct the agent. Every hour you don’t spend fixing hallucinations, architectural regressions, repeated mistakes, or environment issues is an hour the agent stays in flow — and that compounds directly into real productivity multipliers. ...