April 2026
I used AI to build a financial dashboard for my church last week.
Before I did, I spent more time thinking about what NOT to put into it than what to put into it.
If you work in church or nonprofit finance, you already know: we're cautious people. We should be. We're stewards of other people's generosity. Giver data — names, amounts, frequency, contact information — is sacred. Not just legally. Morally.
So when AI tools started showing up in every conversation, my first reaction wasn't excitement. It was: what happens to the data?
Here's what I've learned after months of using AI in my own church finance work:
You don't have to feed it sensitive data to get massive value from it. I depersonalize everything before it touches an AI tool. No names. No individual giving records. Aggregated trends, anonymized cohorts, stripped-down exports. The AI never sees who gave what. It sees patterns, categories, and totals.
The value isn't in the data — it's in the analysis. AI is exceptional at turning a cleaned-up P&L export into a board-ready variance report. At building cash flow forecasts from numbers you've already scrubbed. At drafting the narrative that explains the "so what" alongside the spreadsheet. None of that requires exposing a single giver's identity.
Start with what's already non-sensitive. Your budget is board-approved. Your financial statements are shared with your congregation. Your policies aren't confidential. There's an enormous amount of work AI can help with using information that carries zero privacy risk.
Three practical guardrails I follow:
1. Strip all names and individual records before anything touches an AI tool. Work with totals, tiers, and trends — never raw giver files.
2. Never upload data that could identify a specific person, even indirectly. If a giving tier has only 2 people in it, aggregate it up.
3. Review every AI-generated output before it goes anywhere. AI is a drafting partner, not a decision-maker. You're still the steward.
If you're a church administrator or nonprofit finance leader who's been watching AI from a distance, I get it. I was there too. The question isn't whether to use it — it's how to use it without compromising the trust people placed in us.
That answer is simpler than most people think. And it starts with knowing what to keep out.