5 questions before you start
- What specific task is costing us the most time. Is AI actually suited to it?
- Who will lead this internally, and do they have the time?
- Does the tool we're considering have a clear data privacy policy. Does it meet our standards?
- Do we serve populations where AI errors or AI disclosure would have serious consequences?
- What does success look like at 90 days. How will we measure it?
3 things AI cannot replace
The relationships your staff have built with the people you serve are not automatable. AI can help you communicate more efficiently. It cannot build or repair trust.
AI applies patterns. It does not weigh competing values, consider context, or hold your organization accountable to its mission. That requires a human being.
When something goes wrong, and at some point it will, a human being needs to be able to say "I reviewed this, I approved it, and I take responsibility for it." AI cannot be accountable. Your staff can.
Who needs to be at the table
- Executive Director: sets the mandate, approves the policy, reports to the board
- Internal AI lead: owns implementation, training, and ongoing oversight
- Program staff: know where the time is lost and where the risks are highest
- Development staff: often the first to adopt, need clear funder communication guidance
- At least one board member: governance, not operations, but informed and engaged
The short version of this entire guide
Start with one use case, not five. Pick the task costing your team the most time. Choose a tool with clear data privacy terms. Write a one page policy before you launch. Train staff with hands on time, not a handout. Review every AI output before it leaves the building. Tell your board, your funders, and your community what you're doing and why. Measure what changes at 30, 60, and 90 days. Adjust before you expand.
AI will not save your organization. It will give your team more time to do the work that will. That's a meaningful difference. It's enough of a reason to do this carefully.