Where AI is used — and where it isn't.
We sell AI automation. That obliges us to be exact about our own use of it. This page states what is machine-generated on this website, what is written and reviewed by a human, and what we do inside client projects.
Our AI assistant
The chat assistant in the corner of this site is an automated system, not a person. It says so in its header, in its first message, and in its footer — before you type anything.
- It follows a fixed script written by us. It does not improvise.
- It runs entirely in your browser. Your answers are not transmitted to us, or to anyone else, unless you press send at the end.
- The figure it produces is an estimate calculated from your own four answers — not a measurement, a quotation, or a promise.
- A human replies to every request. The assistant never sends anything on our behalf.
This disclosure is made under Article 50(1) of the EU AI Act, which requires that people are told when they are interacting with an AI system, at the latest at the time of first interaction.
What on this site is AI-generated
| Element | Status |
|---|---|
| Website copy | AI-assisted, human-authored and reviewed. Drafting tools were used. Every sentence was read, edited and approved by a named person, who holds editorial responsibility for it. |
| Journal articles | AI-assisted, human-reviewed. Research and drafting may involve AI tools. Every article is checked and signed off by a human before publication. Any article that is substantially machine-written will say so at the top. |
| Photography | Licensed stock photography. Not AI-generated. If we ever publish a synthetic image, it will carry a visible label. |
| Logo, icons, illustrations | Vector artwork constructed by hand from exact coordinates. Not AI-generated. |
| Code | AI-assisted. Written with AI assistance, reviewed, tested and maintained by a human. Standard practice in software, and we state it rather than hide it. |
| Case studies and numbers | Human only. No figure appears on this site that we have not measured on a live system. Projections are labelled as projections. |
Our labelling rule
We apply a simple standard, stricter than the law requires of us:
- Fully AI-generated content carries a visible AI-generated label at the top of the page or on the image itself.
- AI-assisted content — drafted with AI, edited and approved by a human — is declared here, on this page, rather than labelled item by item.
- Synthetic images, audio or video would carry a persistent visible label. We currently publish none.
Under Article 50(4) of the AI Act, the disclosure duty for AI-generated text applies to text published to inform the public on matters of public interest, and does not apply where the content has undergone genuine human review and a person holds editorial responsibility. Our editorial process meets that condition. We publish this page anyway, because a studio selling AI automation should not have to be asked.
AI in client work
When we build AI systems for clients, the same rules apply — and they are written into the delivery, not left to goodwill:
- Any chatbot or agent we build declares that it is AI at first interaction, in the language of the user.
- Automation runs self-hosted where it matters, so client and end-user data stays on infrastructure the client controls.
- Every workflow is logged and documented, so the client can show a regulator what the system does and on what basis.
- We do not build emotion recognition or biometric categorisation systems, and we do not produce deepfakes.
- AI Act compliance is part of the handover documentation, not an extra.
Models and providers
Where we use third-party AI services in our own work, prompts and content may be processed by those providers. We do not send client personal data to third-party models without a written agreement and a lawful basis. Providers currently in use are listed in the privacy policy.