Soapbox Engage Blog

Most "AI for nonprofits" content this year falls into one of two camps: breathless claims that AI will transform fundraising overnight, or vague caution that doesn't say what's actually changing. What's happening on the ground with development and operations teams is narrower than the hype and more useful than the caution. AI is taking over specific, well-defined tasks, prioritizing a list, drafting a first version, flagging a signal, not replacing the judgment calls sitting on either side of those tasks.

Here's a grounded look at where that shift is actually showing up across fundraising, donor engagement, and the day-to-day operations behind both.

1. Prioritization, Not Prediction, Is Where AI Is Earning Its Keep

The clearest, most reliable use of AI in nonprofit fundraising right now isn't predicting the future, it's ranking the present. That's a meaningfully different claim than "AI knows who will give." It's closer to "AI can tell you who's worth calling first," which is smaller, more defensible, and more actionable, and it's why this use case has moved from pilot projects to standard practice faster than almost anything else in the AI-for-nonprofits conversation.

Predictive scoring models work by pulling together signals already sitting in the CRM and turning a flat list into a ranked one:

• Recency and frequency of past gifts, weighting donors who've given more recently and more consistently higher than a single large gift from years ago.
• Gift size and trajectory, distinguishing a donor whose giving is climbing from one whose giving has plateaued or declined.
• Engagement signals like email opens, event attendance, and website activity, which often surface renewed interest before a gift does.

None of these signals are new. What's new is a model doing the cross-referencing across an entire donor file in seconds instead of a staff member eyeballing a spreadsheet.

2. How AI Is Specifically Changing Fundraising

Fundraising is where most of this is landing first, and it's showing up in three connected ways:

• Ranked donor and prospect lists, built from the giving history, engagement, and gift patterns already sitting in the CRM (the prioritization shift covered above, applied specifically to who gets contacted first).
• Suggested ask amounts grounded in a donor's actual giving history and capacity signals, replacing a flat number applied to everyone regardless of history.
• Channel and timing recommendations that route an appeal to the format and moment a specific donor is most likely to respond to, instead of one send time and channel for the whole list.

None of these three replace a moves-management strategy or a gift officer's relationship. They replace guesswork at the point where a development team decides who to contact, how much to ask for, and when.

The net effect for most teams isn't "AI raises more money by itself." It's that a smaller staff can run a more targeted campaign in the same amount of time, because the ranking, the ask amount, and the send strategy are no longer rebuilt from scratch each cycle.

3. Drafting Work Is Getting Faster, Not More Automated

The single biggest time sink in most development offices isn't deciding what to say, it's getting from a blank page to a first draft. AI has become the fastest way past that blank page, and it's showing up across several kinds of writing:

• Appeal letters and emails that reference a donor's actual giving history and program interests instead of a generic template.
• Thank-you notes tailored to gift size, program designation, and the donor's relationship history, drafted in seconds instead of written from scratch one at a time.
• Grant narrative sections, particularly boilerplate like organizational history or program descriptions, drafted from existing materials and adapted to a funder's specific language.
• Meeting and call follow-ups, turning notes from a donor conversation into a polished summary for the CRM record in a fraction of the time manual write-ups take.

What doesn't change: every one of these is a draft, not a finished piece. Every team using this well still has a person reading, editing, and approving before anything goes out. The time savings come from skipping the blank page, not from skipping the review.

Without AI	With AI, Reviewed by Staff
Starting point	A blank document or a generic template reused for every donor.
First draft	Generated from the donor's actual gift history, program interest, and past communications in seconds.
Staff time	Spent editing, fact-checking, and adding judgment, not typing from scratch.
Before it sends	A human still reads it, confirms every detail is accurate, and approves.

4. Administrative Work Is Quietly Shrinking

Less visible than predictive scoring or AI drafting, but arguably more consequential day to day, is what's happening to routine administrative work. None of it makes headlines the way "AI-powered fundraising" does, but for a stretched two- or three-person development team, this is often the most immediately felt change AI brings to the job.

None of these tasks are glamorous, and none of them were ever going to be the subject of a keynote. But a few reclaimed hours a week, multiplied across a small team and a full year, adds up to real capacity: capacity that tends to get reinvested in the relationship-building work that AI can't do, donor calls, event follow-up, and major gift conversations.

5. Donor Engagement Is Getting More Targeted, Not More Automated-Feeling

Counterintuitively, done well, more AI in the outreach process is producing donor communications that feel less automated, not more. When channel choice, send timing, and message content are all informed by a specific donor's actual history instead of a single segment-wide template, the result reads as more attentive, not less personal. A few examples of what that looks like in practice:

• Channel matching: a donor who's opened the last six emails gets an email appeal; a donor who's never opened one but consistently responds to mail gets a letter instead.
• Send-time optimization: appeals go out when a specific donor has historically engaged, not at the same 9am Tuesday blast for the entire list.
• Cadence control: a donor who already gave to this year's gala appeal doesn't also get the year-end email the following week asking for the same thing again.

The failure mode isn't "AI made this feel robotic." It's skipping the review step and sending an AI draft that references the wrong details. Used as intended, the technology is in service of specificity, not scale for its own sake.

6. None of It Works Without the Data Underneath It

Every shift described above depends on the same precondition: donor data that's accurate, current, and consolidated in one place. A predictive score built from duplicate records ranks noise. A drafting tool referencing stale giving history produces a worse starting point than no draft at all. None of this is a flaw specific to AI, it's the same "garbage in, garbage out" problem fundraising data has always had, just easier to notice now because the tools built on top of it move faster and touch more donors at once.

Before evaluating any AI fundraising tool, a few questions are worth answering first:

We've written in more depth about what that foundation actually requires: Is Your Donor Data Ready for AI-Powered Fundraising? walks through a practical audit before investing in any AI fundraising tool, and Using CRM Data to Find and Prioritize Major Gift Prospects applies the same prioritization logic covered above to major gift work specifically.

Frequently Asked Questions

Is AI replacing nonprofit development staff?

Not based on what we're seeing across development teams. AI is taking over specific, well-defined tasks, ranking a list, drafting a starting point, flagging a signal, while judgment calls like tone, timing, and the final decision to reach out stay with staff.

What's the most reliable current use of AI in nonprofit fundraising?

Prioritization. Using existing CRM data (giving history, engagement, gift patterns) to rank donors or prospects by likelihood to respond is a narrower, more defensible claim than prediction, and it's the use case most teams have moved from pilot to standard practice fastest.

Do nonprofits need new software to start using AI this way?

Not necessarily. Most of what's described here depends on data nonprofits already collect: gift history, event attendance, engagement signals. The requirement is that data lives in one connected system, not a new platform layered on top.

What's the biggest risk in using AI for donor communications?

Sending an AI-drafted message without human review. A draft that references the wrong gift amount, program, or relationship status does more damage than a generic message would. AI works best as a starting point a staff member still edits and approves.

Related reading:

• Is Your Donor Data Ready for AI-Powered Fundraising?
• Using CRM Data to Find and Prioritize Major Gift Prospects
• Fundraising Strategies That Are Working in 2026 for Nonprofits on Salesforce