AI acceptable-use policy
A one-page policy your staff will actually read — written for your sector, signed off at board level, embedded in onboarding. Includes the bits insurers and clients now ask to see.
AI for Business · Consulting
AI for your business — adopted safely, used properly, governed properly.
Microsoft Copilot, ChatGPT, Perplexity and Claude are already in your business — whether you've sanctioned them or not. We help UK small businesses put policy, data classification and the right paid tier in place before the next shadow-IT incident, then build the workflow automations that actually move the needle.
The unsanctioned AI problem
Right now, somewhere in your business, someone is pasting client data into free ChatGPT.
They mean well. They're trying to be efficient. They don't realise that on the free tier, OpenAI may use those prompts to train future models — and that your professional indemnity insurer would like a word.
- Data leakage. Free-tier prompts on consumer AI tools may be used for training. Client names, financials and case details quietly leave the building.
- Hallucinated outputs. Staff paste AI-generated figures into board packs without a sanity check. The model is confident; the number is wrong.
- Shadow IT sprawl. 8 staff, 8 personal accounts on 4 different AI tools, all paid on personal cards. No audit trail, no governance, no shared learning.
- Compliance gaps. Cyber Essentials, ICO data-handling, sector regulators (FCA, SRA, ICAEW) all increasingly want to see a written AI position. "We don't have one" is no longer an answer.
The four platforms we work with
Different tools, different jobs. Pick the right one for each role.
List prices in GBP, exclusive of VAT, indicative for 2026 and subject to change. Get in touch for a current quote against your seat count.
Microsoft 365 Copilot
Inside Word, Excel, Outlook & Teams
From £16.10/user/month
Best for: Teams already standardised on Microsoft 365 who want AI inside the apps they already use, with the security boundary they already trust.
- Email triage and reply drafts in Outlook
- Meeting recap and action items in Teams
- Document drafting from SharePoint content
- Excel formula generation and data analysis
ChatGPT Business
General-purpose AI & custom GPTs
From £16/user/month
Best for: Cross-functional general-purpose AI — sales, marketing, ops, finance — where you want broad use, custom GPTs trained on internal docs, and the strongest model lineup.
- Drafting emails, proposals, briefs, job specs
- Custom GPTs for internal knowledge bases
- Coding help, SQL, regex, ad-hoc analysis
- Image generation for non-brand-critical assets
Perplexity Enterprise Pro
Sourced research with citations
From £32/user/month
Best for: Research-heavy roles — analysts, consultants, journalists, fee earners — who need every answer to come with clickable sources, not vibes.
- Market and competitor monitoring with citations
- Technical research where source quality matters
- Replacing Google for staff who Google all day
- Internal Spaces for shared research projects
Claude Team
Coding & technical reasoning
From £16/user/month
Best for: Technical teams — developers, engineers, technical leads — who use AI as a coding partner. Anthropic's coding benchmarks lead the field as of 2026.
- Pair-programming inside Claude Code (terminal)
- Large codebase analysis and refactoring
- Technical document drafting (RFCs, specs, post-mortems)
- Code review and security-focused diff reading
We'll help you pick the right combination, procure on the right tier, configure SSO and audit logs, and train each cohort. Most clients end up with two or three of these — different tools for different roles.
Free download
The SMB AI Policy Pack — everything you need to ship a written AI position this week.
Most SMBs don't have an AI policy. Most insurers, sector regulators and serious enterprise customers will start asking for one in 2026. Here's the pack we use with our own clients — fill in the form, get the lot in your inbox.
- AI Acceptable Use Policy (editable Word doc)
- 4-tier data classification matrix with worked examples
- Approved-tools register template
- Incident response checklist (data leak via AI)
- Staff one-pager — the do's and don'ts in plain English
Policy pack's on its way.
We've emailed you a copy. To save time, you can also grab it now:
Email not arrived in 5 minutes? Check spam or drop us a line at Contact us.
SecureState™ · Compliance category
Aligned to our IT health benchmark.
Sits inside the SecureState™ Compliance category — GDPR, AI governance, sector regulators and audit readiness. The bit pure cyber audits don’t yet cover, refreshed every quarter.
See how SecureState works
How a typical engagement runs
A predictable shape, scoped to where you are.
-
1
Discovery (1 week)
Half-day on-site or video. Who's already using what, on whose card, with whose data. We map current state, document the leaks, and propose a scope. Discovery is fixed-fee — small enough to say yes to, refunded against the engagement if you proceed.
-
2
Policy & readiness (2–4 weeks)
Draft AUP, data classification matrix, board sign-off. In parallel we run Microsoft Copilot Readiness if M365 is your foundation — identity, permissions, sensitivity labels, Defender. The boring foundational work that everyone wants to skip.
-
3
Pilot rollout (4–6 weeks)
Pick a cohort of 5–15 staff. Procure the right tier on the right platform. Configure SSO, audit logs, conditional access. Run a half-day training session. Measure usage and incidents weekly.
-
4
Scale & automate (ongoing)
Roll out to wider org. Build the workflow automations that emerged from the pilot — usually 2–3 high-value flows. Quarterly review thereafter. Refresh the policy when the platforms change — which they do, monthly.
Consulting, not a package
AI for Business is consulting work — we scope and quote each engagement.
No tiers, no monthly retainer locking you into work you don't need yet. We propose a fixed-fee engagement — usually £2,500–£15,000 depending on size and scope — covering policy, readiness, pilot rollout and handover. Platform licensing is pass-through at list price (we don't mark up your Copilot or ChatGPT seats). Ongoing review is hourly or rolled into your managed service if you have one.
Common questions
AI for Business — frequently asked.
Is it safe for staff to use ChatGPT or Copilot at work?
Yes — but only with the right plan, the right governance and a written acceptable-use policy. Free ChatGPT trains on your data by default; the same prompt run in ChatGPT Business or Enterprise does not. Microsoft 365 Copilot inherits your existing Microsoft compliance boundary. The risk isn't the tool — it's the absence of policy. We help you put both in place before staff make the choice for you.
What's the difference between Copilot, ChatGPT, Perplexity and Claude?
Microsoft Copilot lives inside Word, Excel, Outlook and Teams — best for teams already on Microsoft 365. ChatGPT is the broadest general-purpose AI for drafting, research and custom GPTs. Perplexity is built for sourced research with citations on every answer — strong for analysts and consulting work. Claude excels at long-document analysis and coding tasks. Most clients end up using two or three of them for different roles.
Do you offer AI training for staff?
Yes. We run a half-day on-site or remote workshop tailored to the platforms you've adopted — typically a Copilot prompt-craft session for everyone, plus a deeper session for power users. We follow up with a one-page acceptable-use policy that staff actually read and an internal Q&A channel monitored for the first month.
Can AI replace our admin staff?
No, and we won't sell it that way. AI removes the drudgework around admin — drafting, summarising, formatting, looking things up — but the judgement, relationships and accountability stay human. Clients who use AI well typically grow without growing headcount as fast, rather than cutting jobs.
What is Microsoft Copilot Readiness?
Copilot Readiness is Microsoft's four-stage assessment that checks whether your tenant is in shape to deploy Microsoft 365 Copilot without leaking data. It covers identity hygiene, SharePoint/OneDrive permissions sprawl, sensitivity labelling, and Defender posture. Most SMB tenants fail at least two of the four stages on first inspection. We run the assessment, fix the gaps, then deploy Copilot to the right users with the right guardrails.
How much does AI cost per user?
Indicative list pricing in 2026: Microsoft 365 Copilot from £16.10/user/month (annual, requires M365 Business Standard or higher). ChatGPT Business from £16/user/month annual. Perplexity Enterprise Pro from £32/user/month. Claude Team from £16/user/month annual (5-seat minimum). Add typically £25–£50/user for a structured rollout (policy, training, governance) in the first quarter.
Will AI work with our existing Microsoft 365 tenant?
Usually yes — if your tenant is in good shape. Most M365 tenants we inherit have permissions sprawl on SharePoint, weak sensitivity labelling, and identity hygiene that hasn't been touched in years. Copilot will surface whatever it can see, so any "Everyone" share or orphaned folder becomes a Copilot result. We tidy first, deploy second.
Do you build custom AI agents or workflow bots?
Yes. We build with Microsoft Copilot Studio (great for M365-native flows), Power Automate (for the workflow plumbing), and n8n (for cross-platform automations that aren't a Microsoft fit). Each automation comes with documentation and a handover so your team can maintain it. We don't build agents you can't run without us.
Ready when you are.
Talk to Brett or Simon. 30 minutes, on the phone or video. No deck, no decision pressure — we'll tell you honestly whether we can help.