AI Use Policy

AI Use Policy

The tools we use, how we handle data, who reviews the output — and what we will never do.

Version 3.0 · Last updated: 5 June 2026

(supersedes v2.0 · 5 June 2026 and v1.0 · 24 May 2026)

GridteamAI is a UK & Ireland infrastructure intelligence firm. AI is the tooling we use to scan, summarise and structure the public information our clients need. It is not the thing we sell. This page sets out, in plain UK English, what tools we use, how we handle data, who reviews the output, and what we will never do.

Tools we use

  • Claude (Anthropic) — enterprise tier, no-training settings on. Primary research and drafting tool.
  • Stack AI — orchestration of multi-step research workflows.
  • ChatGPT (OpenAI) — enterprise tier, no-training settings on. Secondary drafting and reasoning.
  • Perplexity — primary source discovery and citation surfacing.

Data we will and will not enter

  • We use enterprise-tier accounts only. Free-tier consumer AI tools are not used for any client engagement.
  • Public UK and Irish government and industry publications are entered freely — they are the foundation of our research.
  • Client material is entered only with explicit written authorisation, and only into enterprise accounts with no-training settings confirmed.
  • Personal data (named individuals' contact details, identifiers, financial or health information) is never entered into any AI tool — for any reason.
  • Classified, security-restricted or commercially-sensitive third-party material is never entered into any AI tool.

Review process

  • Every client-facing output is reviewed line by line by a senior consultant before release. No exceptions.
  • Every cited figure or quote in a published report is traced back to its primary source and the source is listed in the document.
  • Where a source is unclear, contradictory or unavailable, the report says so — it does not fill the gap.
  • For new product lines, full line-by-line review continues for the first twenty deliverables before checklist QC replaces it for routine work.

What we will never do

  • Invent a fact, a figure, a quote, a source or a customer.
  • Publish anything without a named consultant's review.
  • Cite a non-UK publication in a UK-focused intelligence product, unless explicitly relevant and clearly flagged.
  • Use a client's data or our own work to train a third-party AI model.
  • Pretend the AI replaces the expert. The expert remains accountable for what GridteamAI publishes.

Operational standards

  • Open Government Licence v3.0 attribution applied wherever government material is reproduced or quoted.
  • UK data residency for all stored materials, where the platform offers UK hosting.
  • Editorial independence is non-negotiable. We do not accept paid placement, sponsorship, or commercial influence on our published analysis.

Questions or concerns

If a client, partner, regulator or member of the public has a question about how GridteamAI uses AI on a specific engagement or publication, contact book@gridteamai.com. We will respond within one working day.

Version control

  • Version 3.0 · Published 5 June 2026. Compliance claims (ICO registration, Cyber Essentials, PI insurance) removed pending accreditation. Operational standards section added.
  • Version 2.0 · 5 June 2026. Domain updated to gridteamai.com; contact email consolidated to book@gridteamai.com.
  • Version 1.0 · 24 May 2026. Original.
  • Next review: Q4 2026 (or sooner if material change to tooling).

gridteamai.com · book@gridteamai.com