AI Usage Policy
Where artificial intelligence is and is not used in producing this website and in delivering our services. Written plainly, because vagueness here is itself a signal.
Why we publish this at all
A great deal of IT provider content is now generated wholesale by AI and published without anyone technical reading it. It is fluent, confident and frequently wrong — and it is difficult for a reader to tell from the outside.
Since you cannot tell by looking, the only useful thing we can do is state our position and be judged against it.
Where AI is used
We use AI tools the way a competent writer uses a good editor: to check structure, to suggest a clearer phrasing, to catch inconsistency across long documents, and to interrogate a draft for things we have assumed rather than established.
AI has also been used in building this website — in the code, and in drafting and refining content under direction from the business.
Where it is not
No page on this website is AI output published without a person who does this work reading it, checking it against reality, and standing behind it. The judgements — what matters in a medical practice, what actually breaks on a construction site, what a firm should spend money on — come from doing the work, and AI cannot supply them.
We do not use AI to invent examples, to generate statistics, or to write anything presented as direct experience. Where a page describes something as commonly seen, a person has seen it.
The rule that does the work: AI can help us say something more clearly. It cannot decide what is true. Anything it produces is checked against a primary source or removed.
AI and your data
We do not put client data, credentials, network documentation or personal information into public AI tools. That is a rule we apply to ourselves for the same reason we would tell you to apply it: information entered into a consumer AI service has left your control, and no assurance in a settings panel changes that.
Where AI features are built into platforms clients already licence — the assistants inside Microsoft 365, for instance — their use is governed by the client's own tenant configuration and their agreement with that vendor, not by us.
AI in the services we deliver
Some tools we use for monitoring and security detection apply machine learning to identify unusual behaviour. That is ordinary and has been for years; it is how modern endpoint protection works.
What does not happen is a decision affecting your systems being taken by an automated tool without a person reviewing it. An alert is a prompt for a human to look, not an instruction that executes itself.
If this changes
This is a fast-moving area and our position will develop. When it does, this page is updated and dated before the practice changes, not afterwards.