The first time I used AI for a grant proposal, the output came back about sixty-five percent my voice. Not bad. Not good enough. What I did next — and what happened over the following cycles — changed the relationship I had with the tool entirely.
Before I walked into a meeting with a banking executive to make a significant ask, I ran the conversation with AI first. Not to script it — to find the holes. What I discovered in that practice run changed how I showed up in the real one.
I'd known about Community Reinvestment Acts for years. I had relationships in the banking sector. I had a clear sense of what my organization needed. What I didn't have was the bandwidth to think it through. Then I stopped using AI as a search engine and started using it as a thought partner.
After every Defy event, I spent three hours writing individual follow-up emails — different pledges, different conversations, different histories. Twenty-five events a year. Then I spent one Saturday building a workflow that changed that math permanently. This is what that tradeoff actually looked like.
I spent four months integrating AI into 80 percent of my nonprofit operations. When I tried to explain what I had built, people nodded but didn't understand. The moment I showed them, everything changed — and I realized that gap isn't just mine.
I heard it at Harvard Business School, said quietly between sessions: you can make yourself so busy as an executive director that you are no longer accessible to the very people you say you serve. I've watched it happen. And I've come to believe the cost lands not on the leader — but on the people waiting on the other side.
I told a room of nonprofit leaders I work twenty-five to thirty hours a week. The room went quiet. Not the kind of quiet before a question — the kind before a verdict. What happened next said everything about how the sector thinks about sacrifice.
I’ve watched nonprofit leaders wrestle with AI tools built for learners, not doers. The truth? What they really need is AI that learns them — their language, their needs, their pressures — so they can focus on mission, not manuals. Here’s what that looks like in practice.
I’ve seen nonprofit leaders wrestle with AI not because it’s flawed, but because the way we try to use it often misses what we really need. The problem isn’t AI—it’s our approach. Here’s what I’ve learned from the front lines.
Nonprofit leaders aren’t using AI not because they can’t—but because they don’t have the capacity. Here’s what’s really getting in the way.
Most nonprofit leaders don’t have time to learn AI or prompt engineering. This is how AI can actually support real nonprofit work without adding more to your plate.
I started noticing something when I began showing nonprofit leaders AI tools: no one asked to learn how to use AI. They all wanted something that simply worked. This shift in understanding has reshaped how we build technology for nonprofits—tools designed to give leaders their headspace back, not more to learn.