AI on yduman.dev

The Case for Learning Local AI

Tue, 07 Apr 2026 00:00:00 +0000

Note

This post is fully written by me and was checked for typos and grammar errors with Claude.

We talk a lot about agentic coding, but there is a skillset emerging that I think is far more valuable and far less discussed.

I have been deep in agentic coding workflows and debugging with agents. Through that work, I started noticing that the entire conversation around AI assumes that cloud APIs are a given. That assumption is breaking down fast.

My Debugging Workflow with Agents

Mon, 16 Mar 2026 00:00:00 +0000

Note

This post is fully written by me and was checked for typos and grammar errors with Claude.

Debugging is where agents shine a lot. I briefly touched on this in my previous post and I think this topic deserves its own deep dive, because the workflow is more nuanced than just “paste the error and let the agent fix it”. Over months of real-world usage, I developed three levels of debugging that I apply based on how nasty the bug is.

Six Months of Agentic Coding in the Trenches: Lessons from a Brownfield Project

Tue, 09 Dec 2025 00:00:00 +0000

Note

This post is fully written by me and was checked for typos and grammar errors with Claude.

Let’s recap my journey of using various AI tools in a real customer project. The project was going for two years and focused primarily on modernization. What started as legacy hell, is now more maintainable. Deployments are more frequent and the customer gained more confidence developing new features.

The journey is split into three waves of AI usage and the last six months represent the last, third wave, which changed the way I work completely.