How I use AI
I want full transparency in how I use AI tools in my workflow, as well as where I don't.
Where I utilise AI
Research and background reading. Structuring first drafts. Refactoring code. Generation of image and video assets. Speeding up work that is repetitive and would otherwise take a lot longer.
Where AI is not used
My creative vision, design, strategy and final judgement on what I ship to my clients. Everything is read, edited and approved by me.
I don't put confidential information, customer data or commercially sensitive material into AI tools without asking you first.
Provenance and watermarking
From August 2026, AI providers began embedding invisible watermarks in the text and files their tools generate, in order to meet new EU transparency rules. Detection tools are being built alongside them.
This means that if you run a detector across copy or assets on your site, you may get a positive result. It is worth understanding what that does and does not tell you.
A watermark marks text the tool generated. It does not distinguish between work an AI wrote from scratch and work I wrote and asked an AI to tidy, translate or format. The providers say so themselves: a detected mark shows the content was processed, not who authored it.
So a mark is not evidence that nobody was thinking. On any project I deliver, the thinking, the direction and the final judgement are mine.
Where AI generated assets have been used, I keep a record of what was used and where and can share that on request.