Getting started with your AI Hub
This page walks you through both halves of the system: asking questions in the chat, and uploading the documents the chat learns from. Skim what you need, skip what you don’t.
Part 1: Using the chat
This is your daily workhorse. Open it, ask it anything, get on with your day.
What it actually is
It’s easy to think of the chat as a search box for policies — that’s part of what it does, but it’s not the main point. The point is that it knows your organisation. The staff handbook, your policies, your procedures, your service guides, everything you’ve uploaded. All of it sits in the chat’s databank, ready to help you do your actual job faster.
Once you’ve used it for a week or two, most people stop thinking of it as “the chatbot” and start thinking of it as a colleague who has read everything and never forgets. Use it for:
- Drafting emails and letters to service users, volunteers, families, partners
- Writing report sections — board reports, funder updates, annual reviews
- Quick summaries of policies for staff inductions, briefings, or supervision notes
- Checking what your policy says before responding to a query, complaint, or concern
- Drafting standard responses you’d otherwise type out from memory each time
- Helping new staff and volunteers find their feet without bothering colleagues
The chat will pull from your actual documents, cite where each piece came from, and stay in your organisation’s tone if you’ve given it examples in the master script. It’s not generic AI — it’s your AI.
Three real-world examples
Here’s how it looks in practice. Each example shows the kind of thing you’d type, what the chat gives back, and what to do with the answer.
Of course. Here is a draft you can adapt:
Dear [Parent’s name],
Thank you for getting in touch, and for raising your concerns — we always welcome questions like this, because it matters to us that families feel confident in how we look after the young people in our care.
Our safeguarding approach is led by [Named Safeguarding Lead], who all our staff and volunteers are trained to refer concerns to. Every member of the team completes safeguarding training as part of their induction, and refresher training is built into our annual schedule. Our safeguarding policy sets out clear steps for recognising, responding to, and recording any concern…
(I’ve drawn this from the Safeguarding Policy on file. Let me know if you’d like it shorter, more formal, or adapted for a different audience.)
Here’s a draft you can tighten or expand:
“During this quarter the organisation continued to apply its formal complaints procedure as set out in our Complaints Handling Policy. All complaints received were logged, acknowledged within five working days, and investigated by the relevant team lead in line with the policy’s stage-one process. No complaints were escalated to stage two during this period. Trends were reviewed at our monthly operational meetings, with learning fed back into staff training and service delivery. The board can be assured that the procedure is being followed consistently, and that the small number of issues raised have been resolved within published timescales.”
Here you are:
- Treat all personal information as confidential. Anything you see, hear, or are told about a service user, colleague, or volunteer stays within the organisation unless sharing is specifically permitted.
- Know the exceptions. Confidentiality is overridden where there is a safeguarding concern, a legal duty to disclose, or a clear risk of serious harm. In those cases follow the safeguarding procedure.
- Be careful with informal channels. Don’t discuss cases in public spaces, on personal messaging apps, or on social media — even when names aren’t used.
- If in doubt, ask. Your line manager or the safeguarding lead is always the right person to check with before sharing anything.
(Source: Staff Handbook, Section 4 — Confidentiality.)
“What’s our complaints procedure?” · “Who do I contact about safeguarding?” · “What’s the policy on…?” · “Can you summarise the volunteer handbook?” · “Help me write a Welcome message for a new volunteer” · “Draft a thank-you note to a donor”
Personal data about specific staff or service users. Legal advice. Anything not in the documents. If it doesn’t know, it will tell you – it’s designed not to guess.
Walk through it
The interactive demo below shows the chat exactly as your team will see it. Click through the scenes (or let it autoplay) to see how everything works.
How to use the chatbot
Eight short steps showing you exactly what to do. No experience needed.
The “Internet” button — only when you need it
Most of the time, you don’t need this. The chat already knows everything in your uploaded documents and will answer your everyday questions from there. Your policies, your procedures, your handbook, your contacts — all of that is in the databank already.
The Internet button is there for the small number of cases where the answer genuinely isn’t in your documents and needs to come from outside:
- Checking whether a national policy or piece of legislation has changed recently
- Looking up something time-sensitive or very current (today’s guidance from a regulator, a recent news item, something just announced)
- Finding contact details for an external body you don’t have on file
That’s it. Do not use the Internet button for everyday questions — questions about how your organisation works, what your policies say, who does what internally. The chat will answer all of those from your own documents, and the answer will be more accurate than anything the open web can give you.
The blue dot next to the Internet pill stays visibly lit while it’s on, so you can always tell. Tap the pill again to turn it off. When Internet mode is on, answers come from the public web and may not reflect your organisation’s position — always check the source link before relying on the answer, and fall back to your own policies for anything official.
Tips for getting better answers
- Be specific. “What’s the safeguarding lead’s contact?” beats “safeguarding?”
- Ask follow-up questions. The chat remembers context. You can say “and what about for volunteers?” after the first answer.
- Check the citation. Every answer says which document it came from. Click through if you need to verify.
- Use the thumbs. 👍 helps us know what’s working. 👎 flags answers that need improving.
- Switch language if you need to. Top-right corner. 45+ languages supported.
- Voice works too. Tap the microphone to dictate.
Part 2: Uploading documents
For approved staff only. You need a login. This is where the chat’s knowledge comes from.
The two kinds of documents
It helps to think of your documents in two buckets:
Your handbook, your main policies, your volunteer guide. These are stable – they change once a year, maybe less. Upload them once and they power most answers.
Newer policies, updated procedures, board-approved guidance. You’ll add these as they come up. Remove the old version when you upload a new one.
What’s safe to upload
- Published, approved policies and procedures
- Staff and volunteer handbooks
- Service descriptions and FAQs
- Training materials
- Anything that’s already shareable internally
What’s NOT safe to upload
- Documents containing personal data (names, emails, addresses of individuals)
- Draft documents that haven’t been approved yet
- Confidential operational information (rotas, salaries, payroll)
- Meeting minutes, email exports, WhatsApp chats
Walk through it
This second demo shows the upload journey from signing in to seeing your document live in the chat.
How to upload and manage documents
Read the rules first, then the steps. What you upload matters more than how often.
Before you upload anything
Get this right first. Most problems with the assistant come from what was put IN, not the assistant itself.
The do’s
- Do upload documents your team actually gets asked about.
- Do give every document a clear, descriptive title.
- Do keep the most recent version only. Remove old ones.
- Do test with a question after uploading.
The don’ts
- Don’t upload anything with personal data (names, emails, addresses of service users or staff).
- Don’t upload meeting minutes, emails, or WhatsApp exports.
- Don’t upload draft documents that aren’t approved yet.
- Don’t upload scanned images without text (scans need OCR first).
Frequently asked questions
Is the chat private?
Yes. Conversations clear automatically when the visitor closes the tab. Nothing is stored by default. The chat is hosted on UK infrastructure.
What if it gives a wrong answer?
Click 👎 on the answer. We see those, and they help us improve the documents and the assistant’s guidance. The assistant is designed to say “I don’t know” rather than make things up – but it’s not perfect.
Can I add to the documents myself?
If you’ve been given a Document Hub login, yes. If not, ask whoever set the system up. Don’t worry – you can’t break anything. Documents you upload don’t go live until they finish processing.
Can I remove a document I uploaded?
Yes. Sign in to the Document Hub, find the document, click Remove. The chat stops using it immediately.
What languages does it support?
45+ for chat input and output. The documents themselves can stay in English – the chat will translate the answer if you ask in another language.
Does it work on phones?
Yes. The chat works on any modern phone or tablet. The Document Hub also works on mobile but is easier on a laptop.
Disclaimer
Colin will provide the complete disclaimer text. This block holds its place in the page so the layout is finalised. When the text arrives it goes here in place of this placeholder paragraph.
