When it asks if you have any copy or notes, you can upload documents or paste in information to help it understand your work more deeply. AI can only work with what you give it - so if your services are specific, your language is distinctive, or the people you work with have particular needs, share that.

Don't worry if things aren't neatly organised. Bullet points or rough notes, are all fine. Share what you've got.

Important context (add as much of this as you can)

• What your service does - in plain terms, how you'd describe it to a friend in a few minutes, rather than a funder in an extensive report 
• Who it's for - age, circumstances, geography, any eligibility criteria
• How people access it - referral process, self-referral, drop-in, phone, online
• Any practical details - location, times, what to bring, what to expect

Additional context to add depth and detail

• Examples of how you've described your service before - website copy, previous leaflets, induction packs
• Feedback or questions you often hear from new clients - what confuses people, what they wish they'd known
• Your tone of voice or brand guidelines, if you have them
• Translations or Easy Read versions, if relevant
• Any language or terminology you deliberately avoid - and why
  ‍

Before you upload files: check for sensitive data

Review your documents for personal or sensitive information about beneficiaries. Anonymise names and identifying details before uploading. Even though Worthwhile Chat is privacy-first, good data practice starts with you. If you're using other AI tools, check their data settings too - many train on your data by default. How much you share depends on your existing privacy and data policies, and how much control you have over the tools you use.

If you’re unsure, ask yourself: “If this document were seen by someone outside our organisation, could anyone be identified or harmed?” If yes, anonymise first. 

AI will produce a first draft based on the context you've shared. It may also ask you clarifying questions - for example, what happens at a first appointment, or what a specific programme name refers to. Answer these to improve the draft.

Read through what it has produced and ask yourself:

• Would the people we support understand this without having to ask for help?
• Does it answer the most common questions someone new to our service would have?
• Is the tone right - warm and welcoming, not clinical or bureaucratic?
• Are the people we support described in a way they'd be comfortable reading?
• Does it reflect how we actually talk about what we do?
• Are there any assumptions baked in - about digital access, literacy or prior knowledge?
•  Is there anything that could accidentally exclude or put someone off?

If something doesn't feel right, that's fine - you can refine it in the next step.

Getting the right tone for beneficiary communications often takes a few attempts. Think of it as a conversation - keep refining until it sounds like you, and until it would genuinely land with the people you're trying to reach.

Ask for revisions

If the first output isn't quite there, try instructions like:

Tone:

•  "Rewrite this section at a Year 6 reading level - some of the people we work with have low literacy."
• "Make this feel warmer and more welcoming - less like a policy document."

Length:

• "Break this into a simple FAQ format - people often have specific questions they want answered quickly."
•  "Create a shorter version for a text message - maximum 160 characters."

Representation:

• "Rewrite the description of who we support - it feels a bit 'done to', as if people are passive. Can you make it more active and asset-focused?"
•  "Remove any language that labels people by their condition - we prefer person-first language."

Add what only you know

AI works with what you give it. Once you have a solid draft, add:

• Specific local details - addresses, phone numbers, what the room looks like, who will greet them
• The name and voice of a real member of staff or volunteer, if you want the message to come from someone
• Any recent changes to how your service works
• A genuine quote or reassurance from someone who has used the service before (if this is attributed to their name or organisation, ensure you have their consent)

Write for the reader, not the organisation

The most common mistake in beneficiary communications is writing for funders or trustees rather than the people who'll actually read it. If you've been working in your sector for a while, it's easy to forget that terms like "referral pathway", "holistic support" or "lived experience" mean nothing to someone who's never come across them. Ask the AI to flag any jargon it notices - and then trust the feedback.

Think about access, not just language

Plain English is a good start, but it's not the only barrier. Think about whether the format works for your audience. Would an Easy Read version help some people? Does it need to be available in another language? You can ask the AI to produce different versions for different audiences.

Get consent before including quotes

If your draft includes a quote or testimonial from someone who has used your service, make sure you have their explicit consent before publishing. Check they're happy with how they've been represented and whether they want to be named or remain anonymous.

Always fact-check and do a human review

AI can misread details, misname services or fill in gaps with plausible-sounding but inaccurate information. Always check all practical details - phone numbers, addresses, eligibility criteria, opening times - against your original sources before sharing anything. It's also worth testing the draft with someone outside your team, or someone from the community you're trying to reach, before it goes any further. Never share AI-assisted content externally without a human review.

Be transparent about AI use

Consider noting that AI tools were used in drafting this material. More organisations are being open about this - and it's in line with Worthwhile AI values.