How I Use AI
With a book coming out later this year and (at the time of writing) nearly 100,000 words of posts and newsletters on this website, I am getting asked more regularly about my own AI use. It seems to me that writers owe their readers clarity and transparency: obfuscation and lack of provenance trouble me more than AI use per se. If you’re familiar with my work, it should not be a surprise that I take an essentially pragmatic view on the usefulness of AI tools and the validity of AI-assisted content. So, with transparency in mind, this page sets out where I personally use AI in my work and writing. This is a snapshot as of August 2026: it is quite likely that the tools I use will evolve over time, though the underlying principles should be enduring.
Newsletters
To identify potential topics, I use AI research tools including Yutori and Gemini Deep Research alongside news alerts and email subscriptions. The first version of every newsletter is pen on Post-It, mapping out the flow and balance. I write a long-form draft, which is then fed into Claude to give me feedback from perspectives that are representative of my newsletter audience, and to identify any spelling and grammatical issues. It’s generally helpful as a proofreader and less reliable on tone and perspective, so the final version is edited by me and reviewed by colleagues.
The newsletter goes out every Friday. A scheduled Claude workflow extracts the text from the newsletter platform and converts it to Markdown for this website. The following Wednesday, a scheduled task in Claude will pull analytics from my newsletter provider, Kit, via their MCP integration. It also takes data from Google Analytics (integrated via Zapier), and looks through my inbox for any email replies containing qualitative feedback. All of that is put together in a single-page marketing report. Partly that’s about time-saving, as doing all of that work manually would be a solid hour of my time. But more importantly, it drives consistency: in a busy week, a manual reporting task could easily fall down the list.
My Book
My book follows a similar pattern, albeit on a different scale. The ideas, arguments and business examples come from my professional experience. The book was planned offline with Post-Its, my (non-AI) reMarkable tablet and a series of email exchanges with my literary agent. The resulting draft structure was fed into ChatGPT and Claude for early feedback. Once I had written a draft, I used the same models to provide feedback on spelling, grammar and content—alongside more substantive notes from early readers and editors. I also used Claude and Gemini Notebook to analyse the draft and suggest reflective questions, glossary entries and summary tables based on my original writing, all of which I reviewed and rewrote. The book has a transparency statement at the start in line with this page.
My Website
Website content is developed in the same way as newsletters. The structure and appearance of the website were redeveloped using Claude Code. I also use Claude to implement consistent metadata across the site to improve SEO/GEO, and to access Google Analytics data in the same way as for newsletters.
Client Work
I have an early conversation with each new client to determine their approach to AI, with any client parameters reflected in contracts and statements of work. Those arrangements are confidential and bespoke to each client. My approach is always to clearly identify any work that had generative assistance. The core tools that I use in my work are Claude for general analysis, Claude Code for developing scripts and data analysis, and Gemini Deep Research and Notebook for market research and analysis. I also use Claude Cowork and Google Workspace Studio to automate repetitive workflows.
The Bottom Line
If you’ve attended one of my AI workshops, you will have heard me say that AI is a hand, not a brain. My goal is to use appropriate tools for specific tasks, but not to outsource thinking or accountability. I am responsible for everything that goes out with my name attached to it, and if you have questions about any of my work, the best test of its humanity is to have a conversation with me.