My Stance
AI is part of my daily workflow across coding, content, research, and social media. Sometimes it’s a sounding board. Sometimes it’s a drafting partner. Sometimes it’s an agent that can investigate a problem, write code, run tests, or prepare a pull request.
The principle hasn’t changed: AI can act, but I remain accountable.
I review the output and the important actions before anything is merged, published, or sent on my behalf. The agents can do more than they could when I first wrote this page, but they don’t get to make final decisions for me.
How I Use AI
Agents and autonomy
I use multi-agent systems, including agents that can work remotely and delegate tasks to other agents. They may inspect repositories, create branches, write and test code, research documentation, and open pull requests using dedicated accounts and tools.
That autonomy is bounded. I review diffs, test results, claims, and proposed actions before changes are merged or anything is published, sent, or treated as final. Agents aren’t allowed to expose secrets or make irreversible changes just because they were asked to complete a task.
Content
AI may help research, outline, draft, revise, code, test, and fact-check. I remain responsible for the final direction, claims, and decisions. Nothing gets published as my work without my review.
Social Media
Social posts are often drafted or developed by AI from my notes, projects, and ideas. I review the wording, context, links, and claims before posting.
Research
AI is a regular part of how I explore new technologies, understand documentation, and investigate solutions.
Tools
- Claude / Claude Code — Primary AI assistant for development and content
- ChatGPT / Codex — Used for development, research, and content drafting
- Hermes — Harness for my Codex & Claude backed agents
I don't currently use AI image generation tools.
Principles
- AI can assist or act within defined boundaries
- I remain accountable for what gets published, merged, or sent
- All public-facing output is reviewed before publication
- Code changes are reviewed through their proposed changes and verification results
- Agents don’t get unrestricted authority over external systems
- Secrets, credentials, and private material are handled separately from generated output
- I’m transparent about meaningful AI involvement
- AI’s environmental, economic, and creative impacts still matter
This page is inspired by the /ai manifesto. You can find other sites with /ai pages at slashai.page.