How I used AI on this site
The design, the concept and the copy are mine. The build I managed with Claude Code, supervising every step. I take responsibility for everything here.
Mine. The structure of the site, the palette, the typefaces, the visual system, and the editorial decisions about what to say and in what order.
Mine. The case studies, the background and every line on this page come from my own experience and judgement.
Managed with Claude Code. The HTML, CSS and JavaScript were written in that process, under my direction, step by step, with my review at each one.
Mine. I read and approved everything before it went live. If something here is wrong, that is on me and not on the tool.
Why I'm saying so
Because I keep a personal AI policy that requires me to disclose it, always. I use these tools in program management, operations and web development, and I intend to keep doing so.
I also have real concerns about how AI is being used in the world: as spyware, as machinery for misinformation, as infrastructure for concentrating power in the hands of actors who answer to no one. Disclosure is the least I can do to hold both of those things at once.
Where I draw the line
Decisions that affect people: hiring, evaluation, funding, program design. Nor strategic direction, nor client relationships.
Structuring documents from material and judgement I bring, technical tasks with my own review and testing, summarising things I've already read, formatting and proofreading.
Needs active supervision — hallucination is a real and documented risk. If I don't have the domain knowledge to catch errors, I bring in someone who does before that work gets used.
I don't put third parties' personal information, confidential organisational data or sensitive communications into these tools.
The full document
All of this comes from my Personal AI Policy, a living document at version 1.0 since June 2026, which I review every six months. It is grounded in the Framework for AI Fluency (Dakan & Feller) and in Anthropic's guidance for responsible AI use.
I've completed BlueDot Impact's Future of AI and Anthropic's AI Fluency courses. If you'd like the full document, ask me and I'll send it.