💬 AI Tools

How I Work With AI (2026)

My relationship with AI coding tools changed significantly in late 2025 and early 2026. I wrote about this shift in You Are The Driver — the short version is that I now use Claude Code as my primary tool for writing code, while I focus on architecture, validation, and direction.

This isn't "AI writes my code for me" — it's closer to pair programming where I'm the senior dev setting direction and reviewing everything. I've developed a detailed CLAUDE.md configuration that shapes how Claude Code works with me: pushing back on bad ideas, asking clarifying questions, and flagging uncertainty rather than guessing.

• ⚡

  Claude Code — My primary coding tool. I use it for everything from greenfield projects to refactoring legacy code. The key is my CLAUDE.md setup which establishes a collaborative working relationship rather than a "do what I say" dynamic.

• 💬

  Claude — For planning, thinking through architecture, and conversations about code that don't need to touch files directly. Also great for writing documentation and explaining complex systems.

• 🤝

  GitHub Copilot — Still in VSCode for tab completion. I accept maybe 25% of suggestions. It's useful for boilerplate but I don't rely on it for anything significant.

• ✏

  Writing — This site's content is written by me, not AI. Blog posts, opinions, and creative work are human-authored. I'll use AI for editing suggestions or to workshop phrasing, but the ideas and voice are mine.

The Driver Mentality

The framing I use is from You Are The Driver: AI is a powerful tool, but you need to know where you're going. I bring:

• 🎯

  Direction — What are we building? What problem are we solving? What are the constraints?

• 🔍

  Validation — Does this actually work? Does it match the requirements? Is the code quality acceptable?

• ⚙

  Architecture — How does this fit into the larger system? What are the long-term implications?

• ✔

  Judgment — When to push back, when to accept suggestions, when to try a different approach.

See my full CLAUDE.md configuration for how I set up this working relationship.

AI Influence Level (AIL) Scale

I use the AI Influence Level framework to rate content based on how much AI was involved in its creation. Posts on this site may include an AIL rating from 0-5:

• ⓪

  Human Created, No AI Involved — Content created entirely by hand without any AI assistance.

• ①

  Human Created, Minor AI Assistance — Content written by a human, but grammar or sentence structure was improved using AI.

• ②

  Human Created, Major AI Augmentation — Content written by a human but significantly modified or expanded using AI tools.

• ③

  AI Created, Human Full Structure — Human fully described the content, including extensive structure, and AI filled it in.

• ④

  AI Created, Human Basic Idea — Human provided a basic idea and AI implemented it.

• ⑤

  AI Created, Little Human Involvement — AI generated the content with minimal human input or guidance.

Most posts on this site are AIL 0 or AIL 1. When I use AI for code generation or documentation, I always review and customize it to match my standards. You'll find the AIL rating in the metadata of posts where AI assistance was used.