Damion Brown

AI Policy

In a world where we don’t really know whether we’re talking to a human, an artificial intelligence, or a freaky amalgam of the two, I figured it was worthwhile outlining my own personal approach to AI tools and how I use them in my consultancy and advisory work.

I don’t use AI to write or read emails, so everything that comes from me is 100% human, organic, free-range text.

None of the email communications between you and me are stored with or analysed by any AI.

I use an on-device, LLM-powered speech transcriber which is Openwhispr. I use this for voice dictation on my laptop, and occasionally transcribing meetings. No Openwhispr data gets sent to the cloud, it all stays on my device.

On my mobile, I use a similar open source tool called Futo — again, this is completely on-device, with no data sent to any third party cloud service.

I use Google Workspace for my business, mainly because most of the people I work with are in that ecosystem. I don’t use bundled AI services, deactivate what I can, and opt-out of what I can.

I’m not the complete dinosaur that the above makes me sound; AI is probably the most exciting technology I’ve lived through, and not using it would render me pretty useless at advising businesses who operate in a world that’s being upended by that exact same technology.

I use commercial AI tools for:

  • assisting with website maintenance and development stuff (because I’m terrible at it)
  • tuning my LinkedIn posts to match the prevalent trendy format and linguistic conventions on that site
  • para-advisor work, like researching companies, understanding geographic market conditions, maintaining internal notes
  • Brainstorming workflows, frameworks, and analysis ideas that might come in handy in my consulting. I also stress-test certain assumptions about the analytics industry by having AI tools do legwork and research.
  • I use Voicenotes AI to record notes and ideas on-the-go.

As far as is possible, I restrict the ability for commercial AI tools to hoover up data for training purposes, but opting-out of whatever I can, anonymizing anything identifiable, and plain ol’ fashioned using my brain rather than a LLM.

Last updated: 2026-07-15