The Next Nonprofit Software Race Is About Finishing the Work

The most valuable system in 2027 may be the one that helps a team carry a relationship from a signal to a completed next step.

A donor tells a program director why a project matters to her. The conversation lives in an inbox. Six months later, a gift officer is preparing for a meeting with the same donor and has no idea it happened.

The organization has a donor record. It has an email system. It probably has a task list. What it doesn't have is a reliable way to bring the conversation, the person preparing for the meeting, and the next piece of work together.

That gap is where a surprising amount of organizational capacity goes.

The record is only the beginning

For years, software buying has centered on where information lives. Which CRM holds the donor? Which system tracks the campaign? Where do the reports come from?

Those questions still matter. If records are incomplete or unreliable, people cannot make good decisions from them. But a clean database does not, on its own, tell a team who will follow up, what they need to know, or whether the work ever got done.

Consider a donor who gives through a donor-advised fund. The fund sends the gift. A letter names the person who recommended it. Someone needs to connect that information, make sure the right people receive credit, and thank the donor appropriately. If the letter and the gift never meet, the organization can have a perfectly accurate transaction and an incomplete relationship.

The same pattern appears across an organization. A college admissions counselor learns something important about an applicant; a student-support team may need that context later. An event team records attendance; an advancement officer needs to know which conversations deserve a follow-up. A network office needs a view across organizations while each local team keeps ownership of its own work.

The details differ. The question is the same: when something changes, can the right person see it and carry the work forward?

AI needs a place in the work

The conversation about AI often starts with what a tool can generate: an email, a summary, a report. Those are useful outputs. But a polished draft is only as useful as the context behind it and the work that follows it.

An AI teammate can help a team notice that a follow-up is overdue, bring together the information an authorized colleague needs, and prepare a next step for review. It also needs clear permissions, a way to show its sources, and a person who can decide what happens next. An unfinished task should remain visible until someone resolves it or accepts responsibility for the handoff.

That is a different ambition from giving everyone another place to ask questions. It is about helping the organization use what it already knows and making the work easier to carry through.

The demand for a practical path is real. In 2026 research from NTEN and The Bridgespan Group, 70% of surveyed nonprofit leaders and staff said their organizations were missing meaningful opportunities to use AI. Only 8% reported having a one- to two-year AI implementation roadmap. Those responses do not tell us which tools any particular team should buy. They do suggest that interest alone is not a plan.

Start with one handoff

The best first question is not “Where can we add AI?” It is “Where does work regularly stall because the context, decision, and owner are in different places?”

Pick one example your team recognizes. It might be a gift that arrives without enough context for a thank-you, an event conversation that never reaches the relationship owner, or a student who has completed one step but is waiting for help with the next.

Then trace it:

1. What signal tells us something needs attention?
2. What context does the person responsible need—and have permission to see?
3. Who owns the next action?
4. What can be prepared for review, and what decision belongs to a person?
5. What evidence tells us the work was completed?

You may find that the first improvement is a clearer owner or a better handoff, not a new tool. That is useful progress. Once the team understands the work, it can decide where software and AI can help.

A better measure of software

In 2027, I hope we ask less often how many records a system can hold and more often what happens after a person needs help.

Can a fundraiser walk into a meeting with the relevant history? Can an advisor see the next unresolved step? Can a leader find out who owns an issue without chasing three teams? Can the organization tell the difference between work that was assigned and work that was finished?

Software should make the answer easier to find. The relationships still belong to people. The system should help them show up prepared, keep their promises, and follow through.