AI ethics & data handling.

We use AI tools to run our own business — not just to recommend them. Here’s what we use, and why:

• Website chatbot (Google Gemini):the floating chatbot answers FAQs about our service. It only knows what’s in our system prompt — it can’t see your conversation history with us, your bookings, or anything personal.
• Voice agent demo (Retell):the “Talk to Annie” page demonstrates the kind of voice agent we build for clients. Conversations are processed by Retell and the underlying LLM provider; brief transcripts may be stored for quality monitoring.
• Internal research and drafting (Anthropic, OpenAI): we sometimes use general-purpose AI to research tools or draft text. Your specific business information is never put into a public AI model in a way that could become training data.

When you book, pay, and have your discovery call, here’s where your information goes:

• Stripe processes your payment. They receive your billing details. We never see card numbers.
• Cal.com books your call. They receive your name, email, and selected time.
• The discovery callis a one-on-one conversation between you and a human. It’s held over a standard video tool (e.g. Google Meet). It is not recorded or transcribed by an AI service unless you explicitly consent in advance.
• Your written report is drafted by a human using standard productivity tools (Google Docs / Pages). We may use general-purpose AI to help organise or polish wording — your business information is treated as confidential and is not uploaded to any model that uses input data for training.
• Our recordslive in standard business systems (email, accounting software). They’re kept as required by Australian tax law (typically 7 years).

• Use your business data to train AI models.Not ours, not anyone’s. Your information is for delivering your assessment, full stop.
• Sell your data. Ever. To anyone. For any reason.
• Recommend a tool you can’t leave. Every tool we recommend has an export path. You own your data on every platform we suggest.
• Recommend AI for tasks where it’s a bad fit. Some processes are fine manual. Some are dangerous to automate (medical advice, legal advice, decisions with safety implications). We’ll tell you so.
• Hide our use of AI.If a recommendation came partly from AI assistance, we’ll say so. If a tool we recommend uses your data to train, we’ll flag it. Anyone who hides this is being dishonest.

We build AI agents that handle routine, repetitive customer interactions — answering FAQs, taking bookings, qualifying leads. That work is genuinely tedious for humans and AI is genuinely good at it now.

We don’t recommend replacing roles that involve genuine human judgement, empathy, or risk-bearing. AI assistants in those contexts should augment a human, not stand in for one.

For the people you currently employ on the tedious work: well- implemented AI usually frees them up to do the more interesting and higher-paid parts of their job. We see this consistently across our clients.

If you receive a written report from us, you’re receiving a human-authored document. AI may have been used to research tools, check facts, or help with phrasing — but the analysis, recommendations, and judgement are ours. A human will always have reviewed the final document before it lands in your inbox.