AI transparency
Where machine, where human.
An agency selling visibility in AI systems should disclose where it uses AI itself. This page does that, without flattering itself in either direction.
On this website
Use by area.
- Text drafted with AI support, edited and owned by people
- Technical claims from people, checked before publication
- Figures and evidence researched by people, with a source
- Images currently labelled placeholders, no AI images
- Code written with AI support, reviewed and tested
- Replies to enquiries by people only
No text on this website has been taken from a model unreviewed.
In the work
Where AI appears in a project.
In measurement. We put questions to third-party systems and evaluate the answers. That is the core of the service and impossible without those systems.
In evaluation. We use AI to support pattern finding in larger datasets. What a finding means is judged by people.
Not in advice. Recommendations, assessments and anything we assert to you comes from people. A report nobody has read does not go out.
Not with client data without agreement. Which systems get used in your project is in the quote. Confidential material does not go into third-party systems unless explicitly agreed.
Crawlers
How we handle AI crawlers.
We explicitly allow AI crawlers to access this website. The rules sit in the robots.txt, with a separate entry per named service.
That is a deliberate decision rather than an oversight: this website exists to be cited in AI answers. Blocking would be self-defeating.
In addition an ordered short version sits at llms.txt, and every page is also available as Markdown.
None of that applies automatically to client projects. Whether and which crawlers are allowed is the client's decision, and our recommendation differs by goal.
Labelling
Why this is here.
Transparency duties for AI systems have applied at European level since August 2026. Which of them apply concretely to an offer like this is to be settled with qualified advice, and we claim no legal opinion here.
Regardless of the legal position we consider disclosure appropriate. Anyone marketing the idea that AI systems check claims should make their own claims checkable.
This page is updated as soon as the use changes. The review date sits at the top.