How to put AI to work in your charity

A supporter needs a reply. A funding deadline is approaching. There's a trustee meeting to prepare for, and the newsletter still needs writing. Meanwhile, the people your charity supports need you too.

Finding useful AI ideas is one thing. Fitting them into a busy charity is another.

I've already covered ten practical ways charities can use AI. The next step is choosing something manageable and giving your team a clear way to try it.

I'd start with one recurring piece of work, a named person to oversee it and a month to see whether it helps. Here's what that could look like.

Choose the work and boundaries in week one

For this example, imagine a small charity running weekly community sessions. Each month, its coordinator gathers activity figures, meets colleagues to discuss progress and prepares an update for supporters.

That's a useful sequence to test. The team already understands the work and can judge the quality of the results.

Before opening an AI tool, write down how the process works today. Where do the figures come from? Who checks them? Who approves the supporter update? Which parts involve avoidable copying, reformatting or repeated drafting?

Choose a narrow starting point. The coordinator might ask AI to organise an approved activity summary into a meeting briefing. Keep the final decisions and publication with the people already responsible.

Check your organisation's rules, approved tools and relevant data settings. Minimise sensitive information. Use anonymised or fictional examples for early experiments, and remember that removing a name may still leave someone identifiable from the details.

Agree what success would mean. Perhaps the briefing becomes easier to review, preparation takes less effort or colleagues arrive with clearer questions. Include checking and correction time when assessing the result.

Prepare a better meeting in week two

The coordinator can now assemble a small set of approved source material: the previous action list, an aggregate activity summary and any confirmed operational updates.

AI can help organise these into a short briefing and suggested agenda. Give it the meeting's purpose and explain what colleagues need to decide.

Ask it to keep confirmed facts separate from questions. It should flag missing information rather than fill gaps with plausible answers. Request a reference back to the source for each important figure.

Suppose attendance has increased. The briefing could suggest asking whether that reflects additional sessions, more participants or a change in record keeping. It shouldn't confidently choose an explanation without evidence.

This is also where your team's knowledge matters. Perhaps a volunteer has raised a concern that isn't in the figures. Add that context before circulating the briefing.

Read the result alongside the originals. Check whether it's helped colleagues understand the situation, as well as whether the individual facts are right.

After the meeting, use suitable notes - either manually or other AI notetaker - to prepare a draft action list. With Copilot Premium and some other AI notetakers, an action list is automatically recorded. If you need to, manually make clear which points were decisions, which were suggestions and which remain unresolved.

Check owners and deadlines with the people involved. An enthusiastic discussion about a possible event mustn't quietly become a commitment to deliver it.

Create a supporter update in week three

Once the team has agreed what the month's activity means, it has a stronger basis for communicating its impact.

Ask AI to draft a short supporter update from the approved findings. Supply the audience, purpose, preferred length and an example of the charity's writing style. Explain which facts must appear and which details should be excluded.

Review the first draft for substance before polishing the sentences. Does it describe what happened accurately? Does it imply that attendance alone proves a lasting improvement? Has it added praise, quotations or outcomes that nobody supplied?

Then add the human detail you're entitled to share. Check that consent covers the intended use of any personal story. Preserve people's meaning and dignity, including when that produces a quieter story than a fundraising appeal might usually seek.

The approved update could become a newsletter item, a social post and a thank-you message. Review each separately. Shortening a story can remove context that made the original fair and accurate.

A funder may need a different version again. AI can help map the material against reporting questions or grant criteria. Your team must supply the evidence and check the budget narrative. A gap in the evidence should stay visible until it's resolved.

Review the trial in week four

At the end of the trial, compare the process with your usual approach. Look at the finished briefing, action list and supporter update.

Ask the people using them what improved and what caused extra work. Did they trust the source references? Were corrections minor, or did the team have to rebuild the result?

Be honest about an experiment that hasn't helped. A useful outcome might be discovering that AI works well for formatting the action list but needs too much supervision for another part of the process.

Save the successful instructions with a good example. Record the source material required, the checks to complete and the person responsible for approval. Someone else should be able to repeat the task without guessing how it works.

Choose the next task with the same care

Your next experiment might be routine inbox replies, turning notes into a report or reviewing the structure of an existing procedure. Policy drafts still need review by whoever owns the policy, with specialist advice where needed.

Researching funders and partners is another option. Use the Charity Cheat Sheet's FACT check: Find the source, Assess its quality, Compare it with other evidence and Test the impact if it's wrong. Confirm eligibility and deadlines on the funder's own website.

Keep the emphasis on useful work and relationships. Choose one task this week, agree the checks and put a review date in the diary.

Want the full Charity AI Cheat Sheet, or practical training to help your team get started? Email me at jonathan@pollinger.ai or get in touch here and we'll work out a useful place to begin.