AI for nonprofits: where it saves staff time, and what to keep out of it

What can AI do for a nonprofit’s staff?

If the same few people at your nonprofit write the appeal, answer the volunteer emails, update the donor database and prepare the board packet, their hours are the limit. AI helps with the parts of that work that repeat and can be checked:

• Drafting: first drafts of appeals, thank-you letters, newsletters and grant narratives, built from your notes and past materials.
• Summarizing: meeting notes, program reports, survey comments and long funder guidelines, cut down to what matters.
• Rewriting: one update shaped for a donor, a board member and a volunteer, or a dense report turned into plain language.
• Sorting: a shared inbox grouped by request type, and volunteer sign-ups sent to the right coordinator.
• Cleaning records: likely duplicate donor records flagged, addresses standardized, free-text notes turned into fields.

None of this needs a new system on day one. It needs clear instructions, your own source material and a person who reads the result before it leaves the office.

Which AI is best for nonprofits?

Look first at the one inside the tools your staff already open every day, because that’s where your documents and email already live. Nonprofit AI choices fall into four groups:

• If you run on Google Workspace: Google for Nonprofits Help, read in November 2026, says eligible organizations can use the Gemini app at no cost with the Workspace for Nonprofits edition, with more generative AI features in the advanced editions.
• If you run on Microsoft 365: look at Copilot, Microsoft’s assistant, first, for the same reason, and check which Copilot features your plan includes.
• General assistants: ChatGPT and Claude are flexible for drafting and summarizing. They work from what you paste, upload or connect, so their data terms matter as much as their features.
• AI inside your donor database: check whether your donor management system has AI features for drafting, segmenting or flagging lapsed donors before you buy another tool.

Before anyone pastes in real information, read the plan’s data terms: whether what you type is used to train the vendor’s models, who at the vendor can see it, and how long it’s kept. A staff member’s free personal account is the wrong home for donor records, whichever tool it is.

How should a nonprofit use ChatGPT?

ChatGPT for nonprofits works the way it does for any team: a quick first drafter that knows nothing about your organization until you tell it. These habits help:

• Show it your voice: paste in two or three past letters or appeals you like and ask it to match them.
• Supply every fact: program numbers, stories and outcomes come from your records, never from the model. If a draft contains a figure you didn’t give it, take the figure out.
• Ask for options: three subject lines or two openings, and a person picks.
• Save what works: keep the instructions that produce good thank-you letters or grant summaries in a shared library the whole staff can use.
• Use an organization account: so the work stays with the nonprofit when someone leaves, and data settings are managed in one place.

The same habits apply to Claude, Gemini and Copilot.

How can nonprofits use AI for donor engagement?

AI can help with two parts of donor engagement, timing and attention, when it works from your donor database:

• Faster thanks: a personal thank-you drafted from the gift record (amount, designation, giving history) the day the gift arrives, for a person to edit and sign.
• Segments: donors grouped by what they support, how often they give and how recently, so the appeal to program supporters says something different from the one to event guests.
• Notes that turn into follow-ups: a call note becomes a task, a date and a draft reply.
• The right story for the right donor: the person who funds the after-school program hears about the after-school program.

A donor should never receive a message that someone at your organization wouldn’t have written. If the model invents a detail about the donor or about what their gift did, the message loses the one thing it was for.

Does AI improve donor retention?

Not on its own. The Fundraising Effectiveness Project put overall donor retention at 43.3% in 2025, up from 43.1%, in a year when, by its estimates, the number of donors fell 3.6% and dollars rose 5.0%. Its report says the growth came almost entirely from major donors, which leaves the sector leaning on fewer, larger gifts.

What AI can change is the routine work retention depends on: thanking quickly, noticing who has lapsed and following up before the next appeal. A weekly list of lapsed donors with a draft note for each, or a thank-you drafted the day a gift lands, helps keep those jobs on the calendar. Whether that moves your retention rate is something to measure in your own database, a year against the year before.

How do nonprofits use AI for fundraising?

Beyond donor messages, these fundraising uses sit on the grant, research and event side:

• Grant prospecting: a funder’s guidelines, recent grants and stated priorities summarized, so you can decide quickly whether to apply.
• Grant drafts: a first draft of the need statement or program description from your existing materials, shaped to the funder’s questions and word limits.
• Funder reports: program data and staff notes turned into a draft report against the outcomes you promised.
• Events: sponsor letters, run-of-show drafts and thank-you emails to attendees.
• Appeal tests: two versions of a subject line or an opening, sent to matched groups, with your own results deciding.

Read each funder’s guidelines before using AI on an application, and never let a draft claim an outcome your program data doesn’t show.

Can AI write gift acknowledgments and receipts?

It can draft them, but the IRS sets what they must say, and the figures have to come from your gift records. For a gift of $250 or more, the donor needs a written acknowledgment to back up the deduction. The IRS page on written acknowledgments, last reviewed in June 2026, lists what it must contain:

• the organization’s name;
• the amount of a cash gift;
• a description, but not the value, of a non-cash gift;
• a statement that no goods or services were provided, if none were;
• a description and good-faith estimate of the value of any goods or services that were;
• a statement that those goods or services were only intangible religious benefits, if that was the case.

The IRS page on quid pro quo contributions, reviewed the same month, adds a second rule. When a donor pays more than $75 and gets goods or services back, such as a concert ticket, the organization must give a written statement saying that only the amount above the value of what they received is deductible, with a good-faith estimate of that value. The page lists exceptions, such as goods of insubstantial value. The penalty for skipping the statement is $10 per contribution, up to $5,000 per fundraising event or mailing.

That makes receipts a job for automation built on your donor database: the template holds the required wording, the gift record fills in the amount and any benefit value, and AI’s part is the personal note around it. Someone should check the benefit values before an event’s receipts go out.

What should stay out of AI tools?

Some information doesn’t belong in a general AI tool on any plan:

• Whole donor records: giving history, contact details and notes together. Work from the fields a task needs, or from any AI features built into your donor database.
• What the people you serve told you: case notes, intake forms and anything about their health, immigration status, money or family.
• Donors who asked for anonymity: their names stay out of drafts, prompts and shared instruction libraries.
• Passwords, bank details and grant portal logins: never, in any tool.

Then put it in writing. A one-page AI policy that names the approved tools, the data that stays out, who reviews what goes out and how a donor can ask about it gives your staff one set of rules to follow.

Is AI worth it for a nonprofit with a small staff?

It can be, if you start where it costs little and measure what it saves. If you’re on Google’s Workspace for Nonprofits, the Gemini app costs nothing extra to try. Staff time is a cost too: learning the tools, writing good instructions and checking drafts.

A simple test: pick one task you do every week, time it for a month without AI and a month with it, and keep whichever version wins. If AI gives a development director back a few hours a week, that’s time for donors. If it saves nothing after a month of honest use, stop.

Outside help is worth a look once the work has to cross systems, such as a donor database, a giving form and an email tool that need to share records, or when nobody on staff has the hours to set it up. The AI consulting cost page covers what that kind of help costs.

Where should a nonprofit start?

1. Write a one-page AI policy: approved tools, the data that stays out, who reviews what.
2. Check what your tools already include: any AI features in your email, your documents and your donor database.
3. Pick one weekly job: thank-you letters, the board report or the volunteer inbox.
4. Build a shared instruction library: save the prompts that work, each with an example of a good result.
5. Measure for a month, then decide: hours saved, errors caught and anything donors said.

When several jobs compete for first place, the what to automate first guide covers how to choose.

Our AI Consulting starts by finding the hours AI can give back to your staff; then we build what fits, run it and keep it working. Our nonprofits page covers what we would build for nonprofits, and every engagement starts with a 15-minute call.