My AI policy

Written on by j23n in ai

After a lot of back and forth with agentic coding, I've landed on a system that works for me, and it's based on dividing my workload into things that interest me and those I offload to (the ever increasingly capable) AI agents.

Right off the bat, all content on this site is written by me, edited by me and reflect my own thoughts and my own journey.

In projects I'm working on, I try to make the decision how and when to use AI very consciously, which at times is hard because AI is just so good much of the time, and it's just way easier to delegate to it than to learn things myself.

In my projects, I have started adding an AI disclaimer to the end of the README for transparency. It looks like this:

## 🤖 AI Disclaimer

This project uses AI-assisted development tools. See the [AI usage policy](https://j23n.com/public/posts/2026/my-ai-policy) for details.

**Tools**

- Claude Code (Anthropic) · `claude-sonnet-4-6` · Agentic

### Contribution Profile

Phase                               Human│ AI
─────────────────────────────────────────┼───────────────
Requirements & Scope       85%   ████████│░░          15%
Architecture & Design      85%   ████████│░░          15%
Implementation              5%           │░░░░░░░░░░  95%
Testing                   not started
Documentation              20%         ██│░░░░░░░░    80%

**Oversight**: Collaborative
Human and AI co-author decisions; human reviews all output.

### Process

AI agent operated autonomously across multi-step tasks. Human reviewed diffs, resolved conflicts, and approved merges.

### Accountability

The human author(s) are solely responsible for the content, accuracy, and fitness-for-purpose of this project.
---
*Last updated: 2026-02-20 · Generated with [ai-disclaimer](https://github.com/j23n/ai-disclaimer)*

and is actually generated by a tool I wrote (with AI :-))

I try to keep it up to date so it reflects the realities of the project itself.

There is simply no denying that AI is a huge multiplier in my side project work. The concrete interactions with my agents are still evolving but I have two rules:

  1. whatever lands on main is thoroughly reviewed by me
  2. I take every opportunity to learn new things from the AI

You can see some of the interaction patterns in the mosaic repo. In short, I try to use Claude Code as a remote developer, with whom I interact mostly through issues and patches. I assign issues to Claude, it has its own radicle DID and proposes solutions to the issues as patches.

This still requires a significant amount of interaction as I tell Claude to look at issues, reply to comments and submit patches - something I hope to automate soon. The upshot of this architecture is that Claude runs only inside an LXC container, completely isolated from the rest of my machine.