Every playbook here gets run before it gets written up.

The Nonprofit AI Advisor is a weekly publication for people who work inside US nonprofits — directors, fundraisers, comms, ops, program staff. One tested AI playbook at a time.

No vendor pays for coverage and there are no affiliate links. When a tool is good, its limits get written down next to its strengths.

How a tool gets tested

It gets run end to end on real work first — a grant report, a board packet, a batch of thank-yous. Not a demo prompt on a toy input. Then four things get recorded:

• The prompt, verbatim, so you can lift it instead of retyping it.
• Which model, and on which plan. Both change the answer.
• What it cost, and how long it took including the editing.
• Where it broke.

That last one isn't an appendix. It's usually the most useful part, because it's the part vendor documentation never contains.

Getting in touch

Corrections and disagreements are welcome athello@thenonprofitaiadvisor.com. If something here is wrong, saying so is a favor — corrections get published, not quietly edited.