Hi Reader,

When I arrived at Serving Orphans Worldwide, the organization sent four fundraising appeals a year. The one mailed appeal went out at year-end, and the staff stuffed and stamped every envelope by hand around a table.

The board wanted to grow beyond the churches and major donors that had funded the ministry so far. So we changed the rhythm: four emails a month (not all of them fundraising), a mailed appeal every month, and two mid-major proposals a year.

We didn't add anyone to the staff, and we didn't let anyone go. AI is how we did it, because the only other way was spending money we didn't have. New donors went from about 300 in 2024 to about 1,500 in 2025.

I'll be honest about the cost. People went from slightly busy to really busy. But it's good busy, on work that matters, and nobody is working 68-hour weeks.

THE BIG IDEA

For years, fundraisers were told to pick a side. Automation was efficient but cold: a mail merge and a letter that opened with "Dear Friend." Personal connection was warm but didn't scale: you could write about forty handwritten notes a month before you fell over.

What changed isn't that the machine got warm. It didn't. What changed is that the preparation can now be personal without the contact being automated.

That's why I don't think "Should we use AI?" is a useful question anymore. The better question is: where in this relationship does the machine belong, and where does it absolutely not?

I think about it as a trust line.

Above the line is everything the donor experiences as you: the ask, the thank-you, the visit, the condolence note, the prayer, the story of a real person. Below the line is all the preparation that makes those moments possible: research, finding patterns in your giving data, summaries, first drafts, list pulls, visit notes, versioning one appeal for several audiences.

AI works below the line. You work above it.

Your donors draw the line in about the same place. In a March 2026 survey of 821 U.S. donors by BBS & Associates and Campbell Rinker, the AI uses with the most support were back-office work like reducing administrative costs. The most opposed were generated images (25% opposed) and editing stories about the people a nonprofit serves (24%). Donors want their dollar to go further. They also want the moment they hear from you to be real.

There's an honest limit here, too. AI can get you about 80% of a letter in ninety seconds. The last 20% is where the relationship lives: the reason this donor first gave, the detail that isn't in any database field. AI can't supply it, and it doesn't know it's missing. It will hand you a clean, warm, complete-looking letter with a hole in the middle.

Where the line sits is a policy decision, not a technical one. Every organization will draw it a little differently. What matters is that you draw it on purpose and write it down. When there's no line, every staff member draws their own, at 4:30 on a Friday, when they're behind.

ON THE PODCAST 🎙️

We just finished a series on year-end fundraising on the Nonprofit Answers podcast, and we're starting a new one on building your one-year fundraising strategy. You'll discover the secrets of building a plan you can actually execute. Find us on Apple Podcasts, Spotify, Amazon Music, or your favorite podcasting service.

3 THINGS TO TRY

TRY THIS: List ten tasks your team does in a typical week. Sort each one above or below the line. Find the two your team argues about, and decide them out loud this week.

TRY THIS: Before anything goes into an AI tool, strip out names, emails, addresses, and phone numbers, and keep only donor IDs. You keep the key that matches IDs to people. The tool never learns who anybody is.

TRY THIS: Estimate the hours AI saves your team each week, then put a name on them. Write down which donors get a call or a visit with that time. Unnamed, saved time disappears into more email and more meetings.

GO DEEPER

If you want your whole team working from the same line, look at Build Your AI Fundraising Assistant. It helps a small development team build one shared AI assistant that knows your voice, your case, your approved numbers, and your real stories, so everyone starts from the same strong first draft. You'll draw your team's trust line, turn your best appeals into a voice guide, build a fact sheet that stops made-up numbers, and use AI for visit prep without putting donor names or notes into an AI tool. A person stays in the loop on every output. Take a look at the course.​

STEAL THIS

The prompt I'd start with. It takes about twenty minutes, once:

Here are ten pieces of writing I am proud of. Reverse-engineer my voice into a style guide: typical sentence length, how I open, how I handle Scripture, how I handle numbers and data, words and phrases I never use, how I close. Be specific and quote examples from the text. Then list five rules a writer could follow to sound like me.

When to use it: gather eight to ten appeals, emails, or letters you're proud of and paste them in with the prompt. Save what comes back in a project file so every future draft starts from your rules. (If Scripture isn't part of your writing, cut that line.) Then add three phrases you'd never say to a donor as hard rules.

ONE NUMBER

76%: the share of U.S. donors who have used AI tools themselves, according to a March 2026 survey of 821 donors by BBS & Associates and Campbell Rinker.

Nobody is introducing donors to something strange. They use the same tools you do. That's exactly why they'll notice when a letter sounds like it came from one.

WORTH YOUR TIME

​AI Can Draft Your Appeal. It Can't Tell You What Your Donors Care About. A practical workflow for year-end: people write the brief, check every fact, and judge the letter from the donor's point of view. AI handles the drafts and variations.

​AI for Nonprofits, Episode 5: Where AI Actually Helps More from the BBS & Associates donor survey, plus specific uses like prospect research, donor file analysis, and audience versions of one appeal.

​AI for Nonprofits, Episode 8: What AI Cannot Replace The other side of the line: presence, trust built over time, judgment, and care. Good listening before you decide what to automate.

ONE QUESTION

What's one fundraising task you've decided AI will never touch at your organization? Hit reply and tell me.

See you Thursday,

Jeremy Reis

P.S. Build Your AI Fundraising Assistant is built for small development teams and takes about 30 days to set up. You end up with one shared assistant, a written trust line, and a person in the loop on every output. See the details here.​