AI for Charities: Practical Ways Small Teams Can Save Hours Every Week

Ignore the hype and the doom. For small charities, AI is useful in four specific, unglamorous places — and dangerous in a few others. Here's an honest map.

Small charities are being sold two stories about AI: that it will transform everything, and that it will leak your data and make things up. The truth is more useful than either. For a team of two to ten people, AI is genuinely valuable in a handful of specific, unglamorous places — and worth avoiding in a few others.

Where AI genuinely helps today

• Drafting reports and summaries — turning a term of attendance records and session notes into a first-draft funder update. The evidence is yours; the AI just does the assembling.
• Answering questions about your own data — "who hasn't attended in the last month?", "which grants report this quarter?" — instead of building a filter or a pivot table.
• First drafts of routine writing — newsletters, session descriptions, thank-you letters, volunteer role descriptions. You edit; it types.
• Tidying and structuring — turning a photographed paper register or a messy list into structured records you can actually use.

Where to be careful

• Safeguarding decisions — AI can help you find a record; it must never decide whether a concern is serious. That judgement is human, always.
• Anything sent without review — a hallucinated 'fact' in a funder report costs credibility you can't buy back. AI drafts; a person signs off.
• Pasting personal data into consumer chatbots — participants' details entered into a free public tool may be stored or used for training. Use tools that process data within your own workspace under a proper agreement.
• Public-facing chatbots for vulnerable users — the risk profile is entirely different from internal admin. Walk, don't run.

The questions to ask any AI tool

• Does it only see our organisation's data, and is that contractually guaranteed?
• Is our data used to train anyone's models? (The answer should be no.)
• Does it propose changes for a human to confirm, or act silently?
• Is it covered by UK GDPR with a clear data-processing agreement?
• Can we see what it based an answer on?

Start with one workflow

The teams that get real value from AI don't 'adopt AI' — they fix one painful workflow. Usually it's reporting: the hours between "the funder report is due" and "the report is sent" are the most compressible hours in the sector.

Pick that one workflow, run it with AI assistance for a month with human review on everything, and measure the hours saved before expanding.

How JENY approaches this

Ask JENY is the AI assistant inside the JENY impact CRM. It answers questions and drafts summaries strictly from your own organisation's workspace, never sees another organisation's data, doesn't train models on your records, and proposes new records for a person to confirm rather than changing anything silently.

If reporting is the workflow eating your team's evenings, start free and ask JENY your first question about your own data.