Where Do We Start? Answers to the Four AI Questions Nonprofit Leaders Ask Most

Every conversation we have with a nonprofit leader eventually arrives at the same four questions. Organizations and missions vary, yet the questions hold steady.

“Where do we start?”

“We don’t have time to learn it.”

“Are we doing things right?”

“How will I prove it to the board?”

The second arrives as a statement, though we always hear the question mark.

Each comes from a leader who takes stewardship seriously and wants to make a sound decision with limited time. Here is how we respond to each one when it comes up.

“Where do we start?”

Start with what is already happening. In most organizations we meet, AI has been implemented well before anyone made a decision about it. A development associate drafts appeal letters with a chatbot. A program manager summarizes grant reports with a tool she discovered during a webinar. Both acted in good faith, and both made a decision on the organization’s behalf.

Your first step is to find those decisions and bring them into the open. Ask staff, without judgment, where AI already shows up in their work. A short anonymous survey works well. So does a candid staff meeting in which leadership opens by describing its own experiments. People speak freely once they see the executive director has tried the tools, too.

What you learn becomes your starting map. It shows where AI touches sensitive information and where it already saves staff real time. Those two findings tell you where to look first.

Let any new software purchase wait until the map is complete. The tools your staff already use will reveal your real needs more clearly than a vendor demonstration, and your budget stays intact until you know where it belongs.

From there, choose one area that deserves leadership attention now. For a food bank, that may be the data collected at client intake. For a museum, it may be how images of the collection are used and shared. Starting small gives you a decision you can make well and a pattern you can repeat.

Momentum comes from one well-made decision, followed by the next. An organization that settles its first question thoughtfully will find the second easier, because it now has a method. Each decision teaches the team how the organization weighs AI, and that understanding compounds over time.

“We don’t have time to learn it.”

We hear this concern with the most feeling behind it. Leadership needs a working grasp of AI, and a working grasp takes far less time than most leaders expect.

Executive directors already make sound decisions about insurance and employment law without becoming actuaries or attorneys. They learn enough to ask good questions, and they rely on trusted advisors for the rest. AI works the same way. The knowledge that matters at the leadership level fits into a few focused hours: what these tools do well, and what happens to the information your staff puts into them.

Staff who use AI well often recover hours each week from drafting and summarizing. Leaders who invest a few hours in understanding the tools can direct that recovered time toward the mission with confidence. A modest investment at the top makes the savings on the ground safe to keep.

We also recommend learning by doing. Spend thirty minutes using an AI tool on a low-stakes task of your own, such as drafting a board meeting agenda or summarizing a public report. Notice where the result impresses you and where it misses. Firsthand experience teaches faster than any briefing, and it grounds every later conversation with your team.

Finally, share the learning. Ask the staff member who uses AI most to show leadership how she works. A twenty-minute demonstration from someone inside your organization often answers more questions than a full day of outside training, and it signals that curiosity is welcome.

“Are we doing things right?”

Leaders who ask this question are usually further along than they think. The question itself reveals the habit that matters most, which is paying attention.

Four questions tell you a great deal about where your organization stands. First, can you name the person who owns AI decisions? Second, does your staff know which information stays out of AI tools, such as client records and donor details? Third, does a person review AI-assisted work before it reaches the public or a funder? Fourth, have you decided how you will disclose AI’s role in your work when it matters?

A yes to each question means your organization has a foundation to build on. A no simply names your next task.

Doing things right also means aligning AI use with obligations you already carry. Most nonprofits hold commitments to funders and to the communities they serve. A privacy promise in a grant agreement applies to an AI tool as surely as it applies to a filing cabinet. Review the agreements you have signed, and you will often find your first AI guidelines already written in them.

If your organization serves people who have reason to distrust surveillance, the way you use AI with their information is a mission question first and a technical question second. Leaders who frame it that way make decisions their communities can trust.

The work stays current through regular review. Technology changes every few months, so set a standing time, perhaps twice a year, to revisit your answers. Organizations that review on a schedule catch small problems while they are still small.

“How will I prove it to the board?”

Boards ask about AI for good reason. They carry fiduciary duty, and they increasingly see AI on the agendas of peer organizations and in questions from funders.

What a board needs is evidence of judgment. Proof, in this context, means a written record of what the organization has decided and when it will revisit those decisions. We call the distance between what AI can do and what an organization has decided about it the Judgment Gap. Boards rest easier as that gap closes, and a written AI position is the clearest way to close it.

A strong AI position fits on a few pages. It names who owns AI decisions and states the principles that guide them. It also marks the uses that require human review and sets a date to revisit the whole document. Board members can read it in one sitting and refer to it for years.

Bring the board in early. A board that helps shape the principles will defend them later, and a board that reviews a draft arrives at the final vote already informed. Invite one or two members with relevant experience to comment before the full board sees it. Their questions will sharpen the document, and their endorsement will carry weight in the room.

Anticipate the questions board members will bring. Most want to know whether client and donor data is protected and whether the organization could explain an AI-related mistake if one occurred. A good AI position answers both before anyone asks. It shows the board that leadership has thought through the hard cases, and it gives each member language to use when a funder or a reporter raises the subject.

A brief update each quarter, covering new tools adopted and issues reviewed, keeps the board current and keeps leadership accountable. Over time, those updates become a record of steady, responsible stewardship, which is what every board hopes to see.

What the four questions have in common

Each asks, in its own way, for a decision the organization can stand behind. Starting means choosing where to look first. Finding time means deciding how much leadership needs to know. Doing it right means checking practice against principle. Proving it to the board means writing those decisions down.

This work fits a modest budget and a leadership team without technical training. It asks for attention and a willingness to decide.

Your organization already knows how to make hard decisions in service of its mission. AI is one more place to use that skill.

If you would like a clear picture of where your organization stands, start with the Lansby AI Readiness Assessment. It walks you through the five areas we review with every client and shows you where to begin. Consider it your answer to the first question.


Did we use AI for this post? Outlined and dictated by Jennifer De Witt, written, copyedited for grammar, flow, and story by Claude, Anthropic's AI assistant.

Next
Next

What's the True Cost of Sovereign AI?