Transparency

How I use AI

I help businesses use AI responsibly, so it's only fair that I show you how I use it myself. This page covers the tools I use, what I will and won't do with them, and how I build AI for clients.

Last updated: 15 October 2026

My five rules

A person is always responsible. AI drafts. I check, decide and sign off.
Your data stays yours. It never goes into free or personal AI accounts, and it's never used to train AI.
No decisions about people on autopilot. AI doesn't decide anything significant about a person without someone reviewing it.
Tested before it's trusted. Everything I build is tested on real examples, and you're told what it can't do.
I'll tell you when AI did the work. If an AI drafted something important on my behalf, I'll say so.

The tools I use

Claude, by Anthropic: writing, research, and building and testing automations.
ChatGPT, by OpenAI: occasionally, to compare approaches.
Microsoft 365 Copilot and Google Gemini: when I'm working inside a client's own accounts.

I use paid accounts with model training switched off, so my work and my clients' information aren't used to train anyone's AI.

How AI helps me run EmberSeed

Drafting emails, proposals and documents, which I then edit.
Researching options and summarising long material, which I check against the original sources.
Turning my meeting notes into actions.
Building and testing the automations I deliver to clients.

What I don't do

Put a client's confidential or personal information into any AI tool outside that client's own approved accounts.
Present AI output as fact without checking it.
Use AI to imitate a real person, or to write reviews or testimonials.

The one exception is invoicing: I prepare invoices in my own paid Claude account, with training switched off, using only billing details.

How I build AI for clients

In your accounts, under your control. Everything lives in tools you own, and you can switch any of it off.
A person in the loop. Anything that affects customers, staff or money is reviewed by a person before it goes out.
Written down. Every build comes with a one-page guide, plus simple rules for your team about what's safe to share.
Tested. On your real examples before it goes live, with a check-in after a week.
No lock-in. You own what I build, and you don't need me to keep it running.

Following Australian guidance

My approach follows the National AI Centre's Guidance for AI Adoption, published in October 2025. In plain English, its six practices are:

Someone is accountable for each AI tool.
Think through who it affects before you start.
Manage the risks.
Be open about where AI is used.
Test it, and keep checking it.
Keep people in control.

I apply them to my own business and build them into every client project, along with Australian privacy law, including the new rules on automated decisions that start on 10 December 2026. My privacy policy explains how I handle personal information.

Questions

If you'd like to know how AI was used on your project, or anything else about this page, just ask: info.emberseed@gmail.com or 0438 330 100.