AI WITHOUT A WORKFLOW
The executive director has used ChatGPT to draft a thank-you letter. It was fine. She closed the tab. The development director used it once to summarize a meeting, then caught it hallucinating a donor’s name and never went back. The communications associate uses it daily for caption ideas. Nobody has compared notes. There is no shared prompt library, no policy, no agreement about what data goes into a model and what does not. Eight people are using AI eight different ways with eight different outcomes. The organization is using AI. The organization has not built anything with AI.
The 2026 Nonprofit AI Adoption Report from Virtuous and Fundraising.AI surveyed 346 nonprofits and found the headline number: 92 percent are using AI in some form. Only 7 percent report a major organizational or fundraising impact from it. 81 percent use AI individually and ad hoc. Only 4 percent have documented, repeatable workflows. Nearly half have no formal AI policy. The pattern is consistent. The tool is in everyone’s hand. The work that comes out of the tool is not adding up to anything the organization can name.
For a nonprofit at this stage, the conversation almost always starts with the wrong question. The wrong question is “should we be using AI more.” The right question is “what is the one repeating piece of work that AI should be doing for us every week, and what would it take to set that up so it actually runs.” Dan Kershaw, who runs the Furniture Bank and writes publicly about practical AI for nonprofits, frames AI tools as minions. You are the leader. They do nothing useful without clear direction. The framing matters because it puts the discipline back where it belongs: on the human, not the model. For a consultant building AI workflow setup into a client engagement, the same shift: the implementation is not the deliverable, the rhythm to keep using and refining it is.
What Vivia Studios builds inside AI Planning and Workflows is the discipline layer the sector is currently skipping. We start with the one pain point that is most expensive to leave alone. We build one workflow to address it. We document the prompt, the inputs, the expected output, and the human checkpoint. We let it run for a month before adding another. We teach the team how to argue with the model when it is wrong. The Drift workbook, which we publish openly, is one worked example: four domains, one Friday ritual, prompts you can copy and edit for your own practice. The same approach is what we build inside client engagements, scaled to the work the organization actually does. For consulting firms layering AI workflows into nonprofit clients, the same playbook: workflow, documentation, rhythm, teach, hand off.
If you are using AI without a system, or you have written a policy nobody is following, the next move is a thirty-minute conversation about what the one pain point is and what the workflow to address it would look like.
Erin Peshoff is the Chief Curator of Vivia Studios. She has spent thirty years inside nonprofit operations, helped raise over $100 million for institutional missions, and built Vivia around the operating discipline most strategic engagements skip.
SOURCES