Our AI Act compliance.
We sell AI Act audits and operational compliance. We apply the same standard to ourselves: this page documents our own use of AI systems.
The AI systems we use internally
- Anthropic Claude API: content drafting, code generation, internal automation. Hosting in EU regions where the deployment option allows it; zero data retention enabled according to our subscription tier (Bedrock / Vertex addendum).
- Self-hosted n8n (on our VPS): workflow automation; no public LLM call without explicit review by an operator.
- Self-hosted Open WebUI: internal AI interface, sovereign deployment, no routing of data to third parties.
- Public conversational assistant (site chatbot): generative AI operated through providers established in the EU (Mistral AI, Scaleway, France); the only AI system in direct contact with the public.
Risk classification (AI Act, Art. 6 to 9)
None of our systems produce any decision with legal effect concerning EU citizens. Nearly all of our uses are back-office (drafting, internal sorting, document generation) and fall under minimal risk. The only system in contact with the public, the site assistant, falls under limited risk (transparency obligation, Art. 50) and makes no automated decision.
Transparency (AI Act, Art. 50)
Our conversational assistant informs visitors at all times that they are interacting with an artificial intelligence. In addition, if you receive content produced with AI assistance from us, it is reviewed and validated by a senior expert before delivery. We do not deliver any AI output without human supervision. Our methodology is internal and does not constitute a third-party certification (no ISO/IEC 42001 audit at this stage).
Governance contact
AI governance: hello@getulconsulting.com.