Will AI eat the nonprofit world in 2025?

Spoiler: Not completely, but our clients are taking a byte.

 

The dazzling improvement of generative AI platforms in the past year has led to increasing comfort among technologists. 67 percent of tech leaders surveyed by Salesforce reported prioritizing AI, with a third calling it their first priority.

Despite the buzz, nonprofit-specific use cases have been hard to come by—and have mostly been in the direct service realm. At Dreamforce in September, nonprofit leaders touted Agentforce mostly to reroute service calls, which will surely increase capacity for some organizations. However, many nonprofits feel that the technology is not ready for client-facing applications; and one tech leader told us that they are reticent to adopt an agent for something as complex and sensitive as their crisis line.

But what if I told you some of the best nonprofit AI use cases are in operations?

Nonprofit back-office applications for AI

Heller has been working with clients who want to leverage AI tools predominately for operations—accelerating some of the time savings we’ve already seen with automation. Our clients are implementing generative AI for operations in these ways:

1. Precision fundraising: Analyzing giving patterns to identify and surface potential donors who are most likely to contribute.
2. Personalized engagement: An AI can help marketing and communications teams understand constituent behavior and preferences, which allows the teams to tailor messages for the best results.
3. Streamlined reporting: Platforms can autonomously collect, process, and present information, giving leaders real-time data to make decisions.
4. Automation: It can take on routine administrative tasks including data entry, scheduling, and follow-up communications.

Our Microsoft and Blackbaud Practice lead Karen Harris Melendez said Heller clients were already leveraging AI capabilities in Development and Program areas.

For example, she knew a food aid nonprofit that leveraged AI to generate personalized asks via the org’s donation page that was in line with what donors could afford, based on previous giving.  The result was a breathtaking 30 percent jump in donations.

AI also had a critical role in securely analyzing multiple data points already in an org’s CRM and surfacing information, such as high net-worth/high affinity constituents, for planned giving campaigns.

Karen said another use case was in service delivery. A nonprofit working to end food insecurity leveraged AI to optimize staffing and food distribution based on previous outcomes. Their AI-powered systems could predict shifts in need that help them run programs more effectively, she said.

In 2025, nonprofit leaders would be prioritizing the groundwork for AI projects, focusing on the interoperability of their tools and developing data governance that cleared the way for the tools to securely access the institutional knowledge of the whole organization.

Karen, for one, couldn’t be more excited about the applications.

But she stressed that AI was a tool to supplement an organization’s most valued resources—their staff. She said the nonprofit tech leaders she was talking to didn’t want any part in downsizing their teams but, rather, were intrigued at the prospect of leveraging AI to emancipate staff from drudgery, freeing them to do higher-level work.

“You can’t lose the human touch and still achieve your mission,” she said.

How to build your nonprofit AI program

Michael Tjalve, a nonprofit AI specialist, says it is time for tech leaders to get started on a generative AI proof of concept.

In a webinar for threshold.world last month, he outlined his methodology for AI in nonprofit tech:

• Learn the technology. Leaders must understand the capabilities and limitations before asking the team to brainstorm for possible applications.
• Start small. Develop a proof of concept using out-of-the-box tools. Very few nonprofits will need a custom solution right away. Most can use the tools that are already available to figure out and test a use case. Prototype, then iterate.
• Write policy with mission in mind. Your nonprofit’s mission and constituents will determine what policy guardrails are important to you.
• Build an iterative strategy. Borrow best practices from the software development lifecycle to test and iterate multiple times before launching your initiative.

There are a number of responsible AI frameworks. Here’s one from our friends at the Nonprofit Technology Network. Data.org has a range of resources, including a data maturity assessment, policy development toolkits, and training specific to nonprofit tech leaders.

Pitfalls for AI-curious nonprofits

Nonprofit AI pilot projects are not without risk.

Mr Tjalve said AI was, and always would be, imperfect and limited. He named privacy concerns and AI blunders as major reasons nonprofits were slow to get started.

He said leaders needed to develop an AI methodology for their organization that accounted for what he called the “cost of error.” For example, your Netflix recommendations queue is AI-based but the cost of it being wrong about what movie you want to watch next is minimal. By comparison, a mistake made by an AI that helps a humanitarian organization map an area for land mines could be fatal.

The key was for tech leadership to understand what AI platforms could and couldn’t do.

“It starts with building some familiarity with the technology itself, moving away from the hype and what you might have seen from Hollywood and the news,” he said. “(Get) real with understanding what the technology can do, how it works, how it makes decisions and how it makes mistakes.”

But he said nonprofits also can’t afford to let the opportunity go by.

“Start small but do take the first step. Don’t let AI be something that happens to you.”

This blog was written with research assistance from CoPilot and Claude.AI, and transcript support from YouTube, while listening to music suggested by our Spotify algorithm. 🤖