Transparency · show, don't tell
We apply AI governance to ourselves
As the AI Auditor, we hold ourselves to the same standard we audit at our clients. Here is every AI tool we use, what we use it for, and how a qualified human stays accountable for the result. Seven years of hands-on AI work, made checkable.
The tools we use
| Tool | Used for | Governance |
|---|---|---|
| Claude Code | Website & code development | Human review of every change; no confidential client data |
| ChatGPT | Research, first drafts | Every fact independently verified before publishing |
| Gemini | Research, analysis | Cross-checked against primary sources |
| Perplexity | Source research with citations | Sources opened and verified, not taken on trust |
| ElevenLabs | Audio / voice | AI-generated audio is labelled as such |
| HeyGen | Video avatar | Labelled as AI-generated (EU AI Act Art. 50) |
Our rules of use
- Human-in-the-loop: a qualified human (Lars Zimmermann) reviews and is accountable for every published result.
- No confidential client or audit data goes into public AI tools, sensitive work stays in controlled environments.
- AI-generated or AI-modified content is labelled transparently, per EU AI Act Article 50.
- Facts, figures and standards are verified against primary sources before they go live.
- Data protection first: EU-hosted services preferred, data minimised.
Want the same clarity for your AI?
An AI management system (ISO/IEC 42001) gives you exactly this overview, which AI runs where, who is accountable, how outputs are handled.