What Nicolas Rodes Gets Right About ChatGPT for Nonprofits
I use AI in my own work, and the question I come back to is simple. If a draft goes out under my name, can I stand behind every line? That’s what I liked about Nicolas Rodes’s guide for Elevation. He gives nonprofit teams eight prompts they can put to work and makes clear who’s responsible for the finished product. Read “ChatGPT for Nonprofits: An Honest Guide With 8 Reusable Prompts,” by Nicolas Rodes, published October 28, 2026.
The prompt I’d try first
Start with the newsletter example in Rodes’s guide on Elevation. Give ChatGPT a few notes your program team has cleared for public use, and ask it to flag anything that needs checking. I’d take a draft with a hole in it over one that quietly invents an attendance number. The same goes for a grant application. If a funder asks where a figure came from, someone on your team should be able to pull up the source.
As a nonprofit recruiter, I was glad to see Rodes’s job description prompt call for a salary range and ask employers to pare back the requirements. Here’s what I’d add. Before you write the posting, ask what the person actually needs to know on day one and what they can learn after they start. If the role calls for managing a budget, say what they’ll manage. A vague line about “financial acumen” leaves candidates guessing about the job.
Where I’d go further
In his guidance on privacy and board summaries, Rodes tells readers to keep private donor and client information out of prompts. I’d carry that rule right into the board packet example. A single packet can contain a public program update alongside a personnel matter or a confidential financial discussion. I’d pull out the sections approved for use in the tool before asking for a summary. The same rule applies in recruiting. An approved role brief can go in. Candidate résumés and interview notes stay out of the drafting workflow.
My favorite advice in the piece is to put one person in charge of checking the finished work. On a small team, it’s easy to assume somebody else checked the numbers before an email goes out. Pick that person before you start. Try one newsletter item using facts you can verify, and see how much time you actually save after editing. That’s the test I’d use before asking the whole team to adopt it.