Acceptable use of AI.

The short version

We use AI openly, carefully and to your benefit. Four commitments sit underneath everything else on this page:

1. A named person is accountable for every piece of work we deliver. AI is never the author of record and never the excuse.
2. Your confidential information only ever touches enterprise AI tools with contractual no-training guarantees. It is never used to train anyone's models.
3. We will tell you where AI has materially contributed to work you receive.
4. If you have your own AI policy, tell us and we will follow it. If you want an engagement run without AI, we will do that too — just expect it to take longer and cost more.

Why we use it at all

Our view, stated plainly: AI has made the mechanical parts of advisory work fast and cheap. Drafting, summarising, restructuring, first-pass analysis, formatting — work that consultancies used to bill weeks for. We use AI for exactly that work, and we pass the speed on to you.

What AI has not changed is the part you actually hire an advisor for: knowing which question matters, reading a room, weighing evidence, making a recommendation and standing behind it. That is human work. It stays human at Nordant, and it is what you are paying for.

Where we use AI

• Drafting and editing documents, which are then reworked and reviewed by the person responsible for them.
• Synthesis: pulling structure and patterns out of large volumes of material, notes and data.
• Research support: locating sources, summarising public information, preparing background.
• Analysis at pace: building models, tables and first-pass views that a person then interrogates.
• Our own operations: scheduling, admin and internal tooling.

Where we don't

• Recommendations. Every conclusion, score, judgement and recommendation in our work is made by a person who can defend it in the room.
• Anything we sign. Deliverables go out under a named individual's review, and that review is real — line by line, not a skim.
• Sensitive judgement calls: people, organisational and politically sensitive findings are handled by humans, start to finish.
• Fabrication risk zones: we do not use AI-generated facts, figures or citations without verifying them against the source. If we cannot verify a claim, it does not appear in our work.

Your data

• Client information is only processed in enterprise-grade AI tools under contracts that prohibit the provider from training models on it.
• We never put client confidential information into consumer AI tools.
• Access is limited to the engagement team. Client material is returned or destroyed at the close of an engagement on request, AI tool workspaces included.
• Where an engagement involves personal data, our data protection terms apply in full; AI processing does not dilute them.

Accountability

We remain fully responsible for our work regardless of the tools used to produce it. If something we deliver is wrong, that is our problem, not the model's. Our contracts say the same thing this page does: transparency on AI use, enterprise tools only for client data, human review before anything is delivered, and full liability retained by us.

What we ask of clients

Very little. Tell us if you have an AI policy we should follow. Tell us if any material you give us must not touch AI tooling, and we will handle it manually. And if anything on this page prompts a question, ask it — we would rather have the conversation at kick-off than mid-engagement.

Keeping this honest

This policy is reviewed every six months, or sooner when the tools or the law change materially. The version and date at the top tell you what you are reading. Questions to hello@nordant.co.uk.