What I’m Actually Doing With AI in Nonprofit Marketing Right Now

Most of what gets written about AI and nonprofits is either breathless, “this changes everything,” or dismissive, “it can’t replace real connection.” Neither quite matches what it is actually like to use these tools inside a real marketing function day to day, especially right now, when the tools themselves are changing faster than most organizations can keep up with.

Here’s the honest version, as of this moment.

Leading marketing and fundraising for a federation spanning 21 countries and six languages, I use AI tools constantly, including Claude, ChatGPT, marketing automation features, translation support, and the AI functionality built into our existing platforms. Not experimentally. Operationally.

What changed for me was not just that the tools got better. It was how quickly “good enough for a first draft” became “good enough that I have to think harder about what is actually mine.” That line keeps moving. The leaders who will use AI well are not the ones who picked a stack and stopped paying attention. They are the ones still asking, every few months, what just got easier than it was six months ago.

Right now, AI is excellent at speed. A first draft of a donor email, a starting translation before a native speaker reviews it, a way to restructure years of website content during a platform migration instead of reviewing every page by hand. These are real, meaningful time savings, not hypothetical ones. The work does not disappear, but the blank-page problem does, and that matters more than people give it credit for.

Used well, with the right strategy, audience insight, source material, campaign goals, donor context, brand voice, and constraints, it becomes a genuinely useful thought partner. It can pressure-test a message. Surface gaps in an argument. Offer alternate campaign frames. Help clarify whether a donor journey actually makes sense, or challenge a weak call to action. I lean on it this way constantly, and it makes the work better.

But there are a few things it still cannot do on its own, and they are worth naming specifically, because “judgment” alone is too vague to be useful.

It has no stakes. I know what it costs an organization if a campaign misreads a moment, financially, reputationally, or to a donor relationship that took years to build. AI does not carry that risk, so it cannot weigh a bold idea against a safe one the way someone who will actually answer for the outcome can. It will offer a direction with the same confident tone whether that direction is right or quietly catastrophic.

It does not have institutional memory. It does not know we tried this exact campaign angle in 2019 and it fell flat, or why. I can feed it that context, but it does not carry the felt sense of having lived through it, the instinct that kicks in before you can fully articulate the reason something feels wrong.

And it can sharpen conviction, but it cannot originate it. The seed idea, this story, this moment, this ask is the right one, comes from actually caring about the cause and the people behind it. AI is excellent at refining that instinct once I bring it. It does not generate it on its own.

None of that makes it less useful. If anything, naming these gaps clearly is what lets me use AI more, not less, because I know exactly which parts of the work still need to sit with someone who has something real at stake, real history with the organization, and a genuine reason to care how the story turns out.

The strongest leaders in this moment are not the ones who have decided AI changes everything or nothing. They are the ones who can name, specifically, where the line is, and who are still watching it move.