AI Trainer in London: Hands-On AI Training for London Teams, Claude-First

Training only helps if it matches the accounts your people actually have. Here is what the vendors’ own pages said when we checked them in October 2026. Plans change often, so we check again before every engagement.

Claude (Anthropic)

Claude comes in Free, Pro, Max, Team and Enterprise plans. Anthropic lists prices in US dollars. The Team plan adds single sign-on, central billing, admin controls for connectors, shared projects, Claude Code and Cowork, and Anthropic states that it does not train its models on Team content by default. Enterprise adds role-based access, SCIM, audit logs and custom data retention. The Claude add-ins for Excel and PowerPoint are available on the Pro, Max, Team and Enterprise plans, and Anthropic has released Claude for Word in beta on Team and Enterprise plans. Most of our London sessions use Projects, file analysis, the Office add-ins and, for teams on Team or Enterprise, Cowork.

ChatGPT (OpenAI)

OpenAI’s business offer is ChatGPT Business, with Standard and Premium seats. OpenAI says business data is never used to train its models. ChatGPT Enterprise adds SCIM, role-based access controls and data residency options. If your firm has standardised on ChatGPT, we run the same exercises there: projects, file analysis and a team prompt library.

Microsoft Copilot

Microsoft separates Copilot Chat, a secure AI chat with enterprise data protection, from paid Microsoft 365 Copilot, which works inside Word, Excel, PowerPoint, Outlook and Teams using your organisation’s data. Many London firms already have Copilot Chat switched on. We show staff how to use it well, and where a Claude or ChatGPT workspace adds something Copilot does not.

For a side-by-side view, read our comparison of Claude, ChatGPT, Gemini and Copilot.

This is the agenda we start from for a mixed group of non-technical staff. We change the exercises after the scoping call so that every block uses your team’s own work.

A full day adds a function-specific block in the afternoon, for example finance commentary, legal research, campaign planning or customer service replies. It also adds a longer session on building small tools with Claude Artifacts. For engineering teams we add Claude Code in place of the document blocks.

A Holborn law firm: clause comparison without client data

A trainee group takes three anonymised supply agreements from the firm’s precedent bank. In Claude they build a table comparing limitation of liability, termination and governing law clauses, and ask for each clause reference to be quoted. The supervising associate then checks every row against the documents. The exercise teaches two habits: always ask for the source wording next to the summary, and never let the table reach a client until someone has read the clauses themselves. The firm’s ChatGPT users run the same exercise in their own workspace, and the two groups compare results.

A Soho agency: pitch research with a brand voice

A strategy team has five days to prepare a pitch. They create a Claude Project holding the prospect’s public annual report, recent press releases and the agency’s own tone-of-voice guide. Claude drafts a competitor landscape and three campaign territories. The team then works through a claims check, because anything that ends up in an advert has to be backed by evidence. Finally they turn the research into a simple pitch tracker built as a Claude Artifact, which the account lead can update without help from IT.

A London borough or charity: consultation responses

A policy officer has 400 free-text responses to a local consultation, exported with names and contact details removed. In Microsoft 365 Copilot or Claude, depending on what the organisation has approved, she groups the responses into themes, counts each theme and pulls out representative quotes. Before the summary goes to members, she checks the counts against a sample of the original responses. The session follows the Government Digital Service’s AI Playbook and the organisation’s own data protection policy.

Every session includes a data module, because most AI mistakes in UK organisations come from what people paste in, not from the model itself. This is general guidance, not legal advice. Your data protection officer or legal team sets the final rules.

• UK GDPR still applies to AI. If staff put personal data into an AI tool, the organisation needs a lawful basis, has to tell people about the processing, and has to keep the data secure. The Data (Use and Access) Act 2025 has amended parts of UK data protection law, including the rules on automated decision-making, so check the ICO’s current guidance rather than older summaries.
• The ICO has published guidance specific to AI. Its resources include “Artificial intelligence (AI) and data protection”, “Explaining decisions made with AI” and an AI and data protection risk toolkit. We use them to explain fairness, transparency and the data protection impact assessment (DPIA) your organisation may need before a wider rollout.
• The account matters. A personal Free or Pro account is not the same as a business plan with admin controls and contractual terms. We show staff how to tell which one they are using and which one the organisation has approved.
• Leave out what you do not need. Use anonymised or public material for practice. Remove names, account numbers and health details before analysis, and keep special category data out unless it has been explicitly approved.
• A person makes the decision. AI drafts and suggests. Decisions about customers, staff or applicants stay with a person who has read the source material.

Regulated firms add their own layer on top: FCA rules for financial services, professional conduct rules for solicitors and accountants, and the advertising codes for agencies. We build those into the exercises once your compliance team has told us where the lines are.

We quote a fixed fee in pounds sterling for the whole group, after a scoping call. We do not publish a rate card because the fee depends on the number of people, the number of groups, whether we tailor exercises to more than one function, and whether you want a follow-up session.

To help with budgeting, here is what other London providers listed publicly when we checked in October 2026. Open one-day ChatGPT courses started at around £350 plus VAT per person. Private half-day workshops for a team started at around £1,500 plus VAT. Multi-week rollouts and executive programmes cost considerably more. Our list of ChatGPT trainers in London links to each provider’s own pages so you can compare.

What affects the quote

• Headcount, and how many separate groups that needs.
• How much of the material we prepare from your own documents.
• Live online or at your London office.
• Whether you want a review session and champion coaching afterwards.

Whoever you hire, including us, ask these questions before you sign.

1. Will people work on their own tasks? Demonstrations are easy to forget. Ask what share of the session is hands-on and whose documents are used.
2. Which tools, and which plans? A trainer should teach the account your staff will actually use on Monday, not a different tier.
3. How is output checked? Ask to see the review routine. If checking is not taught explicitly, errors will reach clients.
4. What do they say about UK GDPR? The trainer should explain what may be entered, why, and who decides. Vague reassurance is not enough.
5. Can they work with your regulators’ rules? Financial services, legal and public sector teams need exercises that respect FCA, SRA or public sector requirements.
6. What is left behind? Look for written briefs, a team workspace, data rules and a 30-day plan, not just slides.
7. How will you measure it? Agree one or two measures at the start, such as hours on a recurring report or turnaround time for a document.
8. Are the credentials real? Check any claim of vendor certification or partnership on the vendor’s own website. We make no such claim: we are an independent practice.
9. Does the timing fit? Sessions should fit around UK working hours, month-end and client deadlines.