<start of post>

A lot of AI guidance starts with the tool.

Which model should we use? What can we automate? What prompt will get us the best result?

Those are not bad questions. But they are usually not the first questions.

The first question is: what are we actually trying to make easier, better, more accessible, or more human?

Because if we start with the tool, we can end up designing our work around whatever the tool happens to make possible. And that is a strange way to make decisions about mission, staff capacity, community trust, or the people who will live with the consequences of those decisions.

This is especially true for small and midsize organizations. You may not have a full-time technology team. You may be working across programs, fundraising, communications, operations, and direct service with the same five people in every conversation. You may be trying to make thoughtful decisions while also answering emails, meeting with partners, and keeping the actual work moving.

So the goal cannot simply be “use AI more.”

The goal has to be: build enough understanding and enough shared practice that your organization can decide where these tools belong, where they do not belong, and what human judgment needs to remain in the room.

That might mean starting with a low-risk administrative task.

It might mean testing a tool with fictional or public information before bringing any sensitive data into the workflow.

It might mean creating a simple review practice so that no AI-generated output moves directly to a client, donor, community member, or decision-maker without someone checking it.

And it might mean deciding that a particular use case is not worth the tradeoff. That is also a responsible outcome.

I think we sometimes talk about AI literacy as if it is just the ability to write a good prompt. But literacy also means understanding what the system is doing, what it is not doing, what assumptions are built into the workflow, and where the real decision power sits.

Who gets to decide?

Who gets to question the output?

Who gets to say, “This is not working for the people we serve”?

Those are governance questions, even if they never appear in a formal policy.

And honestly, this is why I am less interested in whether an organization has an AI policy than I am in whether people have a way to talk about the choices underneath it.

What information can be used?

What needs consent?

What gets reviewed?

Who is accountable when something goes wrong?

What will the organization do with the time or capacity it hopes to gain?

That last question matters most to me. If AI gives us back time, focus, energy, or capacity, we need to be intentional about where those resources go. Otherwise, efficiency just becomes another way to ask already-stretched people to produce more.

So yes, learn the tools.

But also learn your own values, your own constraints, your own people, and your own responsibilities.

That is where the real work starts.

If your organization is beginning this conversation, I made a free guide with 10 grounding questions and concrete next steps. It is designed to help you get started without pretending that one template can answer every question for every mission-driven organization.

You can find it here: https://example.org/ground-your-ai-guidelines

And if you are already having these conversations, I would love to hear what questions are coming up for you.

<end of post>