AI in Fundraising

Where AI earns its keep in development work — and the lines a fundraiser with integrity does not cross.

Four high-leverage uses

• Prospect research: synthesising public information on a foundation's strategy, recent grants, and program officer priorities into a one-page brief.
• Appeal drafting: generating three tonal variants of the same story so you can pick the one that sounds like you — not the one the model prefers.
• Segmentation: clustering donors by giving pattern and interest, then writing distinct messages for each — instead of one bland appeal to everyone.
• Stewardship: turning impact data into personalised thank-you notes that mention the specific program a donor supported.

The dignity test for appeals

AI defaults to poverty-porn framing because that's what it was trained on. Before an appeal goes out, apply the three-question filter from Module 3 · Discernment:

• Is it true? If a statistic has no source, remove it.
• Is it ours? If it sounds like every other NGO appeal, rewrite it.
• Does it dignify? Would the person in this story sign it if you showed them?

Lines not to cross

• Do not paste donor CRM exports into public chatbots — POPIA and donor trust both fail here.
• Do not write fake beneficiary quotes. Real ones, with consent, or none.
• Do not auto-send AI-drafted messages to major donors. A model cannot hold a relationship.
• Do not use AI-generated photos of "beneficiaries." Ever.

The one-hour weekly ritual

Book an hour a week where you and the model draft the next appeal together — with your voice brief pasted at the top of the conversation. That habit, repeated for a quarter, will move your fundraising output more than any tool purchase.

Continue: AI for Nonprofits South Africa · AI Tools for Grantmakers