H Claude at HTBG

Welcome to Claude

Your new AI colleague across Alpha International, HTB, St Mellitus and Revitalise Trust. This guide takes about 10 minutes — read it, try a few prompts, and tick off the checklist as you go. Your progress is saved on this device.

0% of your first-week checklist complete
01 · Basics

What is Claude, actually?

Claude is an AI assistant from Anthropic. Think of it as a very capable, very fast colleague who's great at drafting, summarising, explaining and organising — but who still needs you to check its work.

Chat mode

A conversation window — ask questions, get drafts, brainstorm, learn something. Nothing on your computer is touched.

Cowork mode

Claude can work with files in a folder you choose, run tasks over several steps, and build documents, spreadsheets or trackers for you.

Skills

Pre-built playbooks for specific jobs — writing a status report, building a spreadsheet, drafting a Word doc. Claude picks the right one automatically.

Artifacts

Living pages Claude can build for you — like this guide — that you can reopen any time and that can refresh themselves with new information.

02 · House Rules

The HTBG AI Use Policy, in plain English

This is the short version of the organisation's AI policy. It applies whichever entity you work for — Alpha, HTB, St Mellitus or Revitalise. When in doubt, ask your line manager or the Service Desk before pasting something in.

What you must not put into Claude
Unless your specific use case has been approved, don't paste in:
  • Personal or special-category data on staff, volunteers, congregants, donors, beneficiaries, students or suppliers
  • Safeguarding records or disclosures
  • HR, disciplinary, sickness or recruitment material
  • Legal advice, privileged material or live dispute content
  • Confidential financial, banking or unpublished budget/pricing data
  • Sensitive IP, strategy or partner-confidential material
Who's accountable for the output
Everything Claude produces is a draft, not an authority. You own accuracy, bias, tone and confidentiality checks. Never let Claude's output be the sole basis for a legal, HR, safeguarding, finance, disciplinary or pastoral decision — those always need human judgement.
What's fine to use day-to-day
Low-risk internal work is welcome: drafting and rewording, summarising and structuring documents, brainstorming, condensing notes, extracting meeting actions, explaining code or writing documentation, and internal ideation or mock-ups that aren't for publication.
Publishing anything externally
Anything meant for the outside world — web, social, press, ministry materials, campaigns — is internal-only by default. It can only go external once the use case is approved, a human has done editorial/brand review, and copyright, licensing and reputational/doctrinal fit have been checked.
New tools and paid use
Don't connect unapproved tools or personal/consumer accounts, and don't treat an embedded AI feature in another app as pre-approved. Paid, pilot or strategic use needs a documented use case, a named business owner and system owner, a funding source, and review dates.
If sensitive data has already been entered: treat it as a potential data-protection/security incident and report it to the Service Desk immediately. Don't wait to be asked.

Quick check — is this OK to ask Claude?

03 · Try It

Prompts worth stealing

Copy any of these into a chat with Claude and adapt it. All of these sit comfortably inside the "everyday internal work" the policy welcomes.

Everyday admin

Summarise this email thread into three bullet points and tell me what needs a reply.
Turn these messy meeting notes into a tidy list of decisions and action owners.
Rewrite this message so it sounds warmer but still professional.

Comms & creative (internal draft — human review before anything goes external)

Give me three different opening lines for a newsletter about our upcoming event.
Brainstorm 10 icebreaker questions for a team away-day.
Suggest 5 taglines for a volunteer recruitment drive — playful tone.

Learning & prep

Explain [a policy/process] to me like I'm new to the organisation.
Help me prepare talking points for a meeting about [topic].

Ops & IT flavour

Turn this jumble of device/security alerts into a plain-English handover note.
Explain what this spreadsheet formula is doing, step by step.
04 · Toolkit

What's already connected

Depending on your role, Claude may already be able to read from these tools when you give it permission. If you need something else connected, it needs a documented use case and a named owner first — talk to IT.

Connected

Microsoft 365

Search across Outlook email/calendar, Teams chats and SharePoint — handy for "find that document" or "what did we agree on".

Connected

Jira & Confluence

Look up tickets and project pages, draft updates, and keep documentation in sync.

Connected

Canva

Get design feedback, make edits, or check a design against brand guidelines — still needs a human sign-off before anything is published.

Needs approval

Anything else

Other tools aren't switched on by default. If your role needs something new connected, raise it with IT with a documented use case, business owner and system owner attached — never connect a personal or consumer account yourself.

Connecting a tool means Claude can read from it on your behalf — always check what you're granting access to, and never use a personal/consumer account to do it.
05 · Your First Week

Checklist

Tick these off as you go. Saved automatically on this device.

You're all set — nice work! You've covered the basics, the house rules, and had a go yourself. Go find something real to try it on.
Finished? Let your line manager or the IT team know you've completed this guide, so we know who's up and running.