Beyond “Using AI”: Predictive, Generative, and Agentic in Fundraising

Speaker 1 0:07
Good afternoon, my name is Matt McAdams, and I am a DonorPerfect training specialist. Welcome to Scott’s session: Beyond Using AI, Predictive, Generative, and Agentic in Fundraising. Scott Rosencrans is head of AI growth at Evertrue, where he helps nonprofits apply Read More

Speaker 1 0:07
Good afternoon, my name is Matt McAdams, and I am a DonorPerfect training specialist. Welcome to Scott’s session: Beyond Using AI, Predictive, Generative, and Agentic in Fundraising. Scott Rosencrans is head of AI growth at Evertrue, where he helps nonprofits apply AI and analytics to strengthen donor engagement and drive mission impact. With nearly a year decade of experience designing predictive and generative AI products for the nonprofit sector, Scott specializes in translating complex technology into strategy, training, and services that deepen, not replace, human relationships. He’s a certified AI governance professional, co-host of the Fundraising AI Podcast, and co-author of Nonprofit AI: A Comprehensive Guide to Implementing Artificial Intelligence for social good. Here’s Scott Rosencrans.

Speaker 2 1:04
Thank you, Matt, and thanks everyone for joining. I’ve been doing the DonorPerfect conference a number of times now, and I always like coming to this community. People just motivated to do good in the world and to do it as efficiently as possible. So, I’m excited to be talking to you all today about beyond using AI. I’ve been in again the nonprofit sector for 20 years, been leveraging AI and fundraising for the past 10 years, and I know that even though the technology has evolved drastically, the conversations have evolved just as much on what is it, how do we use it? How should we use it, but also, how do we get the value out of it? So, I really want to break down the different tools, help shed some light on how the tools are designed differently, and the purposes and objectives that they each have, so you know when you have a certain question that you’re trying to answer, what tool is best designed to help answer that question. So, we’ll go through that today, but before we dig in, a little bit about Evertrue. You might be more familiar with the donor search side. Donor search has been around for 15 years, but they, we got acquired by Evertrue in September of last year, and so now coming together as a wholly cohesive unit. But with Evertrue, we believe fundraising has always been about relationships. It’s not just about the tools, the technology, but the tools and technology are designed to give you as a human more time to engage in human relationships, right. And so we want to make sure that what we’re providing to our organizations, to our clients, to our partners are impactful as possible. Everything we build is designed specifically for mission-driven organizations, profits, healthcare foundations, educational institutions, and our role, our goal is to help fundraising teams turn information into action by knowing who to prioritize, how to engage, and what to do next, answering all the questions and helping you use the data to do your work as efficiently as possible, and again get that time back, so you can spend it more in that human to human element. When teams have clarity momentum, they spend less time managing the systems and more time advancing the mission that matters now more than ever. And we’re excited to be a part of that journey with a number of you here joining today. For more than 15 years, we partnered with organizations navigating the same challenge of rising expectations, increasing goals, but fewer resources every day to do the work that you need to do, being asked to personalize more, move faster, steward better, uncover new opportunities without adding any headcount. So that’s why innovation is so important. It’s not just about the tools, but it’s about what we can do with the technology, with the data, with AI, and how do we make sure that we’re meeting you where you’re at and helping you just move a little bit further down your goal or down the road of executing on your mission. So, more than 15,000 organizations trust our solutions, because we meet practical ways to create more impact across multiple teams and with DonorPerfect, donor CRM to integrate donor searches AI predictive tools or enhanced core scores. So, if you’re a DonorPerfect client and you’re leveraging this data, wonderful. I’d love to hear how it’s working. If you aren’t leveraging this data, hopefully this conversation can help you a little bit further understand what it is that we’re building, and if you don’t have access to it, reach out to your DonorPerfect rep and they can get you set up with more information. And then, lastly, on the Evertrue side, we really don’t think about fundraising in silos, everyone on your team at the organization, not just in your department, is all working towards the same goal, right? And so we want to make sure that, again, we’re meeting you where you’re at by telling you, giving you intelligence information to identify who to reach out to and when and what to do with them, giving the tools to actually do the outreach, so you could focus on the messaging itself, and then stewarding the.

Speaker 2 4:59
As individuals, so that they know exactly where their dollars are going, where they are in that relationship with you, because trust is such an important element in nonprofit fundraising, and so we want to make sure that we’re being helping you be transparent in that case to help instill confidence and trust in the donors that that you’re showing up how you say you’re going to show up, so now moving in, AI is anywhere and everywhere, right? It’s hard to turn on the news, it’s hard to talk to a colleague, it’s hard to talk to your children without hearing about what AI is and what it could do, and all the great things, or all the horrible things that are going to happen, like it’s just everywhere. It is the buzzword of, you know, the last five years for sure, but going on the next decade, because it’s just constantly advancing and becoming faster and can do more and more expensive and all these different things, but it’s just like everyone is talking about AI, but the problem is they’re not really talking about, like, what is it, right? There’s AI in your toothbrushes and your vacuums, and like every single thing that you’re doing now, but like what is it? And it’s so hard to wrap your hands around it because it means a lot of different things. The term artificial intelligence was coined 70 years ago, and it was actually coined by the Rockefeller Foundation in a grant asking for research to study technology and computers to see how they can be designed to execute in a way that humans execute, but over the past 70 years, what AI can do has clearly evolved. Back then, a sophisticated AI model might have been building or playing a really incredible game of checkers, right, but now, like, that’s laughable. Now it can navigate the English language far better than I can, and I’ve been using this language for quite some time. So it’s hard to wrap your hands around all these different things, because AI is just this umbrella term, but there’s so many distinct and different tools within this field that it seems it’s just like nebulous. It’s hard to like really wrap your hands around it, hard to really like make sense of all the things, because it seems like you can do anything and everything. How is that possible coming from just one technology? And so we want to break out some of the, or the three different tools that are really most impactful for nonprofit fundraising, and try to get away from conflating them into just saying, “Oh, well, use AI for that, or “We’ll use AI, and have you started using it? Like AI itself is not a solution to your problems, but there are capabilities within different types of AI that can help you solve some of the challenges that you’re facing as a nonprofit organization. So, predictive, generative, and agentic AI, they are not interchangeable terms. They do very different jobs, they carry very different risks, and they require different adoption strategies all of them, no matter how sophisticated they are. They all come down to you doing the human work. These tools are just surfacing the data, taking the information, giving you what you need, so that you can have more of that face-to-face time, while this manages a lot of the data flow, the decision making, and so on, behind the behind the scenes. So let’s kind of unpack these a little bit. Very high level behavior and find underlying pattern type of AI has been used for a while now, it’s been around for well over a decade in fundraising, specifically, but it’s focused on a predefined set of, sorry, I just got a notification, predefined set of answers that it could provide for predetermined questions, it’s very objective, it’s very logical, it’s very structured, generative AI, the one that’s more attractive, kind of more the buzz. When people are talking about AI now, they’re probably talking about generative AI. This is the Chachi BT, the claw, the open claw, all these different things. This really popped into the scene in November of 2022 when Chachi PT became public to anybody, anybody with a website and internet access had access to this generative AI tool of it, which was not as sophisticated back then as it is now, but still very sophisticated for its time.

Speaker 2 9:34
Generative AI synthesizes vast amounts of data to compile information in new ways. This is more open-ended, more subjective, more creative. You ask the same question a number of different times, and even if you were at the same way, the answers you get back are a little bit different. So, it’s more fluid than predictive AI. Predictive AI is very structured. Agentic AI is the newer one on the scene. Authentic AI really kind of started in 2025 in terms of like the public space and how people can use it, but 2026 is where you’re hearing a lot more about it, and by the end of this year it’s going to be much more prevalent, much more accessible, and also much more sophisticated, just because of how artificial intelligence evolves and is able to just be more impactful, just be more efficient, and constantly with so many resources behind it, with data science teams building out new tools and trying to get more out of it, like we’re seeing the massive increase in what agentic AI can do, although these are three different tools, and again, we’ll get into more detail in each one specifically, but the best organizations and the best value is when you put all three together, because they do different things, they work well together, they’re not duplicative, they’re not redundant, but they help continue a process, and the way I like to really think about this is, my, my wife, my son, and I are actually going on a road trip today, but the analogy that I like to think of is as a road trip, right? Predictive AI would be the first step. I say we’re going to Albany, New York, from Warwick, New York, and that is my destination, that is my goal. Predictive AI would figure out all the routes, look at all the available routes to me, take into account the weather, the traffic, the time that I’m leaving, tolls, all these different things, and say, okay, this is your route, and this is how long it’s gonna take. That’s what predictive AI does, is predicting that journey for me when I give it the end goal. It’s very structured. Generative AI, the next step is that turn by turn navigation as I’m in the car, if I make a wrong turn, which embarrassingly happens way too often when I’m following GPS. I don’t know why they can get confused when there’s all these different exits. Anyway, that’s a neat thing. GPS adapts to where I’m going, adapts to if I made a wrong turn or if I made a pit stop, and it recalibrates to then say, okay, based on where you are now, here’s how you’re going to go. It’s more responsive, more reactive to any information you give it to get you to whatever the end goal is that you’re looking for. And then, lastly, agentic AI – I’m not using this in my road trip – but agentic AI is more like the self-driving car, you’ve given it the destination, you’ve told it how you want it to get there, and it’s executing on that for you. You could sit back, you could sleep eventually, right? But these are all different tools that work together to get you to your end result quicker, more efficiently, sometimes better than we could do as humans, but they are all tackling a different thing again. It’s just when they work together, you’re getting the most bang for your buck by being able to leverage these different types of technologies, and so it’s not really about choosing one and just saying, well, I’m not going to use the others. It’s about choosing which one for your specific needs, right? There’s so many different challenges that we as nonprofits are facing every day. Again, I started as a prospect researcher. I’m very familiar with the challenges that you’re facing every day, and so these tools are designed to help address those in different ways. So, let’s dig more into predictive AI. This again is the more matured of the three. I’ve been building and overseeing predictive AI models for nonprofits since 2017 Back then, conversation about AI was completely different than it is now. People told us, you know, it was science fiction, and they said that we were doing a disservice to the sector by promoting this type of stuff, like snake oil.

Speaker 2 13:35
But predictive AI really is a powerful workhorse by taking so much information and finding like reverse engineering patterns and behaviors and characteristics of how people are engaged are going to engage in the future. It really helps you make the decision of who to prioritize way better than any human could. Again, I started as a prospect researcher. You lock me in a room with a database, it would take me, you know, weeks to figure out who the top, like, 10 people are, and even then I wouldn’t be sophisticated enough to have dug deep into those patterns to really say these are the people that should be prioritized. We, as humans, are only able to assess so much data at a time, but predictive AI is able to take 1000s if not millions of data points and really base off decision trees that get into layers way lower than we would be able to even come up or formulate in our minds, so these types of tools, they’re although they have been around the longest, although they are the most impactful, they’re really underutilized. Only 2.3% of nonprofits are using this to find mid-level and major donors. I do think the reason for this is because since predictive AI is harder to grasp because it’s built behind. In the scenes, it’s harder to like interact with, and you just get like an output in a data file with scores, like it’s not as exciting or attractive, and harder to implement and realize, because it’s not doing work for you, it’s just giving you information to help you do your work better, right? So 10 years in, only 2% of nonprofits are using this type of tool, and really, when we think about the challenges that we’re facing as a sector altogether, the one that concerns me the most is what we refer to as the generosity crisis. My partner in all things AI, Nathan Chatel, he wrote the book Generosity Crisis, November 2022 right, when ChatGPT came out, and it’s really talking about how we’re seeing a significant decline in donors, fewer donors, fewer dollars, pretty much year after year. Donor participation has collapsed from 65% of all US households to down to 45% in two decades, and that number is on a consistent trend down to zero in the next 40 years, and then donor numbers dropped 3% in 2025 which is the fifth consecutive year of decline. Predicted AI is designed to address the generosity crisis. The generosity crisis is here because we are using maybe the wrong data, the wrong tools, the wrong methods to identify the people that we’re going to prioritize, and so when gift officers are calling people in their portfolio, oftentimes those phone calls don’t get answered, the emails don’t get responded to, the meetings don’t get had, and the people that are raising their hand, saying, “Hey, I really love what you’re doing, I really want to support you more, they get overlooked because they don’t prioritize efficiently or effectively. Predictive AI kind of throws that whole model on its head and helps identify those indicators of the people who are raising their hands, saying that they want to do more with you, saying that they want to have a strong relationship, saying that they really believe in your mission, that often get overlooked. They can get prioritized with predictive AI, so until we put the right people in the right place, we’re going to keep facing the generosity crisis. We’re going to keep seeing fewer donors, fewer dollars, because everybody’s just chasing wealth, and wealth is not why people give. When we talk about what this means in fundraising, particularly I again, when it’s set out with a specific target, it can effectively, it can execute on that target, but the target needs to be clearly identified. So, when I work with organizations on building, we build custom AI models at Donor Search. Any model can be built, any target, as long as it meets three criteria. It has to. There has to be enough people in your database that have done x, whatever that thing is. Oftentimes, it’s make a gift next year, make a major gift. Who’s going to increase their giving? What’s the likelihood that they’re going to lapse?

Speaker 2 17:52
Whatever it is, whatever that x is, there have to be enough people that have done it within a recent time period for us to find enough patterns to reverse engineer number one, number two is you need to have enough data that leads up to that activity, right. And so, if there’s, if we just know that somebody made a gift, but we don’t know anything about them before, it’s going to be hard to do those reverse engineering of those patterns, right. So, we need to have data leading up to that activity, and then the target that they did that we said there needs to be enough people on it needs to be clearly defined within the data. There needs to be a way for us to clearly target and say this person did x, because we can see within the data they did x. Once we have that, we have enough people, we have enough historical data, anything can be done. The most often models that are being built are things such as donor scoring, just getting every single person in their universe, your entire universe, and saying here’s a likelihood that they’re going to make a gift in the next 12 months, or likelihood that they’re going to be acquired in the near future, and figuring out the patterns that show their high levels of engagement and high likelihood that they’re going to do that future activity. Some organizations are using this for major gift potential. Major gift potential for organizations are all different, right? Somebody, $100,000 might be a major gift, $1,000 might be a major gift. So, tailoring this to your specific needs, as long as you meet those three criteria, you can use this for major gift potential. Upgrade, upgrade is a one that we’ve been building for a while that I really like, because there are tons of people who are not just on board with your mission, but wanting to do more, right, showing up in certain ways more than they were before. These people, if identified correctly, could be moved from annual giving to leadership annual giving, or leadership manual giving to major gifts, major gift to principal gift, whatever it is, but you need to find those patterns that, that help identify that they’re likely to increase their giving, so that’s what upgrade potential is. Acquisition, just like it sounds, the thing that I think about when it comes to acquisition, if ever, if any of you have ever used an acquisition model in the past. You’ve likely been using a model that says which of your non-donors look like your current donors. The problem with this is your non-donors have don’t have the same opportunities to engage as your current donors do. Your current donors are invited to different events, they have different communication methods with internal team members at your organization, they know about priorities or projects coming up before the rest of the community does, so we can’t say these non-donors look like your current donors, because the same doors are open to them. When we build an acquisition model, what we’re saying is, Who have you acquired in the near future, and what activity led up to that, or first gift? And then, how do we find other people who are engaging in the same way and showing their interest in the same way to those that you recently acquired, playing giving potential, that’s a key one, finding the people that are flying under the radar consistently, but showing up over and over and over again, and meet those criteria again, there’s so many targets that can be used for predictive AI, as long as it meets certain criteria of enough people have done it. It’s been tracked in the data, and there’s just historical data leading up to that activity, but these are common ones that you’re probably most likely to see, and ones that you might be able to have access to with the enhanced core scores from Donor Search within the DonorPerfect product. When it comes to results, these results are significant. We have worked with organizations where they’ve seen they’ve raised $2 million from a single donor identified by this tool that would have often been overlooked.

Speaker 2 21:29
Actually, I should have updated the second one, this $50,000 one, because the organization found someone who was giving $5,000 to that organization but gave significant gifts to other organizations, and had a very high predictive score. They originally gave $50,000 but increased their giving to $75,000 It turned a $5,000 donor into a $75,000 donor, and closed that in two weeks, which is significant in its own, and it’s all because the patterns were there to identify that this person is interested and involved and invested, but the typical data points weren’t showing that they should be prioritized above others. We also have organizations who are decreasing the number of people that they’re mailing to, but seeing higher levels of revenue from those mailers, so being more targeted but getting better results, because they’re reaching out to the people that are interested and want to give, and not just going to throw that direct mail piece away, and then increasing gift size. These are the types of tools, or the results that are really honing in on that generosity crisis, finding the right people, finding the needles in the haystack, getting in front of them, and I, and putting off to the side, really, that 95 to 99% of your database that just right now is not likely to make a gift in the near future, but oftentimes get prioritized for the wrong reasons. So that’s where predictive AI really comes in to shuffle the deck of your database and prioritize those who need to be prioritized. Moving on to the next tool, generative AI. This again is the more recent, the not the most recent, but more recent, as of November 2022 So it’s what, three and a half years old now, going on four years old. And this is the one that when people talk about AI, they’re probably talking about generative AI. These are large language models, so it’s looking at all the text on the internet, all I mean, vast amounts of data, all coming through text or visualizations. Generative AI can also create videos or images, and so on, but it’s synthesizing all that information and packaging up in new ways. Now, there’s there’s thought about can AI create new ideas, like it’s not necessarily creating new ideas in its current state, but it might be creating new ideas to you, ideas that you’ve never heard of, or you haven’t thought of. That doesn’t mean that no one else has, but it’s surfacing all that information because it has access to so much information, and it’s packaging, packaging up in a way that is beneficial to you, if you’re an auditory learner, or you can listen to a podcast to synthesize information. If you’re a visual learner, it can put together an infographic to synthesize information. If you like bullet points, or narrative, or analogies, whatever it is, you say, “Hey, can you give me this information in this way? And it’ll meet you exactly where you’re at again, it’s very creative, very free flowing, very subjective, and so a lot of organizations are leaning into this when doing surveys, trying to see who’s using AI. 86% of nonprofits are exploring generative AI. It’s also very accessible. You could go from not using generative AI right now to then going to chat.com or claude.com running these websites and using it for free or paying $20 a month, so there’s very little hurdle, I mean the technical hurdle in using this data, there are obviously hurdles in terms of compliance or your organization. Governance policies or your organization’s belief on using AI, or even some, you know, moral questions that come up, because generative AI can be detrimental to the environment from some reports, like, so, but in terms of actually just using the tool, you can go to a website and start using tool today, so the accessibility is very much there, right, so when we think about 2% on the predictive side of using 2% of nonprofits using predictive AI versus 86% of nonprofits using generative AI, and this tool’s only been around for three years, like that is a drastic difference, where generative AI is going to make the biggest difference is in burnout. There’s a survey that comes out every year for the past three years. It’s the Social Impact Satisfaction Report, and it asks nonprofit employees how they feel about their job.

Speaker 2 25:52
In the most recent report, 75% of nonprofit employees are looking to leave their job, 65 are considering leaving the sector altogether, and 60% are over. The reason is that 60% of them are overworked and under-resourced. That was the number one reason. Predictive AI is not going to help this. Generative AI helps us. Generative AI helps you do more things with your time, because it can draft those emails, which is a common example. It can put together reports in a way after a meeting, or synthesize minutes that could be delivered to someone, or it can watch an hour long video and give you the five bullet points in just a matter of seconds, so it gives you a lot of that time back, so you can spend your time on that human to human element and not have to get as mired in the administrative tasks or those roles that you aren’t as thrilled about, or that get in the way of you really like leaning into your passion of the work that you do. So, when it comes to fundraising, some examples are crafting content, crafting marketing letters. Now, organizations are finding that a lot of AI is starting to sound very similar, so it can do this, but you’re going to want to have a watchful eye on it to make sure that it doesn’t sound too generic, that it still speaks the story of your organization and has that mission underlined in it, and doesn’t just sound like any other letter getting sent out. More than that, you can tailor them to the individuals themselves, so you have a number of donors in your database, but they don’t all show up the same way, they don’t all the same history, they don’t all the same reasons for giving, but oftentimes they’re all getting the same letter and the same messaging, and so now you can craft a message that’s more specific to them, meeting them where they’re at, letting them know that they’re seen by the organization again, keeping a human in the, in the realm to review these letters to make sure that they are tailored and not generic, but that they are speaking to that person, so that the person feels seen and trusted by you in the work that you’re doing. Donor stewardship also personalized thank you notes. Now, again, at certain levels, you might want those to be actual human-driven thank you notes, but oftentimes the majority of your donors get overlooked, don’t get thanked on time, so being able to send a message out to them in a way that again sees them as a unique individual and unique supporter of your mission, generative AI can help with that tool, taking all the quantitative data they’re giving support and looking at the annual report to say what you’ve been able to execute as an organization, tailoring that to person A to say, hey, thank you for your first gift of $500 the other day, like this really helped us towards whatever, but being able to meet them where they’re at and show them that they’re valuable to your team and that they’re appreciated. Prospect summaries again, starting as a prospect researcher. If I had access to generative AI, my role would look completely different. Generative AI is not perfect; it’s gathering a lot of information, but it gathers that information so quickly that most of my time was spent on looking at different sources and websites and trying to compile this large report on a person, whereas now it could be spent on reading a report that’s been created and just verifying the things that need to be verified, or excluding whatever is not verified. So I can go through a lot more reports than I would have been able to originally, and get back to the work of creating the story of why this person will be a meaningful partner in our mission. Grant writing, analyzing. I never was involved in grant writing, but I’ve seen the process many, many times. And you, as an organization, have one story, have one annual report, have one like this, the list of stats of how you’ve been able to execute in the great work that you’re doing, but organization A or Foundation A wants it in 15 bullets, Foundation B wants it in seven paragraphs, Foundation C wants it in an infographic, like all these different things.

Speaker 2 29:52
I would never want to have to recraft everything, and generative AI can do that so well, and so being able to. Get your time back to be able to address any of these grants and submit applications for these grants, moving much faster than we will be able to do as humans. Social media, same thing, you’re seeing probably a pattern here that generative AI is really great in content creation, putting stories together, creating stories, but it’s also needs to have that human element in it. But this is a great use case for generative AI results that we’re seeing when organizations are using this are 60% productivity improvement for organizations when assessing how productive they were before, how many things that were able to execute on before versus now, like they can do more of the stuff that they want to do more of the tasks they needed to get to, they can focus on them because they can offload a lot of the administrative stuff that generative value can do so well. Reduction time in proposal writing, 50 to 70% increase in grant success rates, because they can really speak to how the foundation is looking to who to deliver grants and the impact that they’re trying to make and connect those dots between what your organization does and what that foundation is trying to do, and then increase in email click-through rates too. Again, like these tools have access to many different writing methods and techniques to make it more appealing. I’m not much of a writer, so generative AI has helped me make my writing a little bit more appealing for putting out on LinkedIn, and so on, and I am seeing that response as a result. It was just up to me, like I’m much better with numbers than I am with words, so I lean more towards generative AI helping me just kind of hone, but it’s always my message coming through, and it’s always my idea that I’m trying to just polish. I’m not saying AI write something for me from scratch and send it off as it’s my own, because that’s not me being genuine or authentic. So, these things are really great in what they’re able to do, but again, they shouldn’t replace or be used to replace the human element of the work that we’re doing. They should be used to augment and allow us to focus on that human face-to-face connection, and then lastly, moving into agentic AI. This, like we mentioned, this is the fully self-driving right, the autonomous vehicles that are able to drive from point A to point B. They’re able to do work, right. So this is more, instead of craft an email, or, you know, write a prospect briefing, it can go into your database again with certain permissions and approvals, and all that stuff, but hypothetically it could go into your database, surface people that have a high predictive score, see what the last time they made a gift was, do research on them, create that proposal, send that proposal to the portfolio manager, the gift officer, and say this is the last time this person had a contact report. Here’s what you need to know about them. They’re very highly scored. Here’s how you should craft the message, and we’ve drafted an email for you. So, like, it is taking multiple steps throughout a process, throughout a workflow, and executing on each one in continuity, so it’s continuing that workflow from start to finish, and just prioritizing where the human comes in to kind of put the final stamp on, okay, this is what we’re doing. It can lower results, it can track data, it can find the throughput through different tools by by connecting into your systems and seeing where people are tracked, but different data sets are kind of like tracking them differently across the organization, and pull all that together to have a better understanding of who these people are, and then execute on that ask. There’s a lot of things that Agentic AI can do, and we’re still very early on on its capabilities. Right now, it’s really great at navigating websites and navigating platforms, anything that could be done digitally on a computer, agentic AI can probably do well now, but will most definitely be able to do very successfully and quickly in the near future.

Speaker 2 33:53
A key part of this again is just making sure that humans stay involved, while it’s easy to turn things over to agentic AI, because it’s, it is so sophisticated. We need to make sure that we are still taking accountability, and we’re still taking responsibility, because agentic AI is not accountable for its work, right? And so we can’t say if something goes horribly wrong, we can’t say, “Oh, well, it was the agent that did it, and you know, so no, no problems here. Like, everything’s okay. It ultimately falls back on us, especially when it comes to relationships with our donors in the community and those that are putting their trust in us. So, really, when we think about agentic AI, it’s proactive. It can take an initiative on its own. It’s multi-step. It can go through a series of steps and execute, oftentimes all at the same time, it uses tools on its own, and it’s goal-directed. You tell it what you want it to do, and it’ll work through every process that it needs to work through to get to that ultimate goal. Again, still being very early, it really kind of launched publicly in 2025 actually December. Of 2025 was when Claude Code and Claude Cowork came out. These were tools that really moved the needle in terms of agentic and being able to do a lot of this work. So it’s effectively for the public sake only seven months old, and we’re still in the very infantile stages of this. We will see this to continue to move faster and quicker, and be able to do more over the next coming months, but only 1.2% of nonprofits currently are using Agent AI. I think a lot of this is not just the accessibility, but it’s also knowing that there’s a lot of risk associated with this, because it is taking actions, hesitation, appropriate hesitation, there in terms of we’re not going to let it run wild, and we’re going to be very thoughtful and mindful about where we’re going to apply this type of technology before just dropping in and saying, all right, have at it, but the answer, the question rather, that agentic AI addresses a predictive addresses generosity crisis by identifying the right people for you approach if generative addresses the burnout crisis by giving your team the tools to do more work at the same time while focusing on the work that they want to do. Agent really addresses the staffing challenges. 81% of nonprofits struggle to raise enough cost to cover their costs, so enough dollars to cover their costs in 2025 and so they’re ending years with operating deficits. Right, there’s layoffs. We’ve seen funding challenges, and not even just organizations that were receiving funding from the government, but organizations that weren’t receiving funding from the government are impacted as well, because those organizations that were now are directing towards other resources like foundations that they weren’t originally, and those foundations are giving there where they weren’t giving, or they were giving to other nonprofits before. So we’re seeing a trickle down effect of organizations just from the funding challenges, but again, this is something that’s been going on for a while, with fewer donors, fewer dollars every year, this constant staffing challenge of open roles that just are unable to meet, you know, where there’s such a concern about overhead for nonprofits that we ask for people to do a lot of work again, people that are here in the organizations, the people you attending this, you’re all passionate about the work you do, that passion leads through, but in reality, you’re probably wearing 17 different hats, right, and your role has probably evolved drastically in terms of responsibility over the times that you’ve been at the organization, and maybe it’s you’re being asked to do more with less, like a lot of organizations that we see are, and so staffing is a significant concern for nonprofits. I think from early last year, there were 25,000 nonprofit staff laid off. More than one in 10 organizations have 20% of staff positions vacant. So, like, this is a concern. Agentic AI can address this. Doesn’t mean agentic AI can replace a person, but it can do more than just one task at a time, like generative AI. Generative AI can do really one task at a time.

Speaker 2 38:04
Agent AI can do more than one task at a time. So, where you might need, have needed five people to execute on this one job, maybe you just need three and 20 agents or something like that. So, like, you could just do more with less, covering a lot more responsibility and task again with the human still involved, so this allows to offset a little bit some of those staffing challenges, and we’ll be able to see more as we move on, but the org structure of nonprofits is going to be evolve, evolving just as technology evolves, where it’s not going to be you’re managing a team of three humans, you might be managing a team of three humans and 15 agents, and they’re all designed for their own specific thing, but as they work together, they can help augment the work that you as humans are doing. So, what does this mean for fundraising again? Prospect research workflows, seg doing the segmentation, doing the prospect research, putting that data exactly where it needs to go, and creating the content that could be used for, or the strategy to help identify how to open that door with that individual. Right, that’s more than just creating a prospect profile. It’s multiple steps, all connected throughout from one end to the other, to give the end user exactly what they need, exactly when they need it to be able to just focus on that human element, donor journey orchestration again. Your donors are different, right? They are not all the same. They don’t all fall into the same buckets, and so wanting to identify them, see them for who they are, meet them where they’re at, and make sure that you’re contacting them in a way that is responsive and appropriate and applicable to that person, right. If I’m calling and calling and calling, and you’re just sending me text message and text message, text message, you’re not seeing me for who I am, right. And if the gift officer isn’t aware that I need to be called or have it on their calendar, or they forget, like, then I’m going to fall through the cracks, and I’m going to find another nonprofit to do. My hard-earned dollars towards so being able to adapt those donor journeys in a way that’s responsive and reflective of that individual, but also leverages the multi-channel opportunities that we have. Grant reporting again, seeing which organizations you receive grants from, identifying throughout the financials where those dollars have gone, putting it into a report to show the outcome and the effectiveness of the dollars received to be able to go back and submit another grant successfully in the future, and then multi-channel campaigns as well, again sending out to text message, email, video campaigns, whatever it is, like crafting all the data to put it all together and executing on it, and then tracking it back in the CRM of who got what, so that way the task itself is complete, not just that one role, but everything associated with it. A lot of the administrative tools, tasks associated with the work that you need to do to manage these relationships, they can get automated in a way that is helping you just get your time back at Evertrude. Specifically, we’re building agentic AI into our software as well. So, we’ll be dealing in your future, that’ll automate one to many omni-channel outreach and giving fundraisers more time to focus on those relationships. So, stay tuned. There’s gonna be a lot of rollouts for a lot of organizations that are using and leveraging this type of technology, we’ll kind of get back to that at a certain point, because I think even though there’s a lot of places to use AI, that doesn’t mean that we should use AI everywhere we can, we need to be mindful of what it’s doing for us and organizations and our donors and use it where it’s appropriate, but again, you probably picked up, keep that human element, so get results significant increase in ROI by organizations that are using agentic AI, because they can now do more with less.

Speaker 2 41:56
There’s I don’t know if any of you follow Jensen Huang, who’s the CEO of Nvidia, Nvidia is the company that builds the chips that all these AI tools are leveraging, and he commented on how Meta and AWS and all these different companies are letting go, have significant layoffs of their organizations, like Salesforce, letting off a lot of people, he says that it’s kind of a lazy approach. They’re saying that they could do more with less, and so they don’t need the staff to do all the work. But why not do more with more? Why not let all those people who are there, who are working there, do more work that you couldn’t even do before? A lot of what’s capable with AI is to do the things that you are doing, but do it faster, better, and more accurate in some cases, or do things that you are never able to do on your own, right? Just like be completely creative and innovative, and think, like, what could we do if we had 20 more staff, or 30 more staff, without needing the overhead of 20 or 30 more staff, right? Like, let the tools kind of explore in those areas to say, like, how would we approach this? I see a lot of feedback here, like, this is a really exciting time for us to be able to lean in in ways, and for-profits are doing this very well. Granted, they have significant resources that we don’t have, they also don’t have the same risks that we have as nonprofits, but they are nimble, they’re able to do things, we’re seeing it’s not going to be long until we see a billion dollar company built by one person, one person employing a billion dollar company, because they’re leveraging it to AI, and so companies that are starting now are not building out their processes or their business plans in ways that they would have 510 years ago, they’re thinking completely differently in an AI native mindset. So we can start doing that as well, seeing what they’re experiencing, what they’re doing, and seeing how we can mirror that to organization, allow us to do more with less or more with what we have to take away some of those 17 hats that everybody is wearing, so significant cost reduction per task versus human execution, and also a 10 hour save per staff member every week, that’s 25% your week. What would you do with another 25% The big concern is if you keep going back to doing the admin stuff, that sending more emails, doing those things, like you’re burying yourself in what you don’t want to be doing. Think about what makes you passionate about the work. Why are you here? What got you excited the day that you got the job, and like you ran and told everybody, “Hey, I’m going to be working at so and so as such and such, right? Think about that, lean into that, and use the 25% of your week that you just got back to lean in there as much as possible, right. That’s going to help us with a burnout crisis, help us again engage with individuals in a way that they’re going to want to support us. So, like, it helps a lot. It has a snowball effect, even though it’s not directly addressing all those things, but it’s us making the decision of where are we going to prioritize that new. Found time that we just came across, so putting it all together. Those are the three tools we’re just going to do, like a quick snapshot of, like, how each of these address the certain, the same elements, and what it, what the outcome is. It’s just kind of like a summary or a recap. So, predictive forecast outcomes, it looks at what’s happened in the past to determine what’s going to happen in the future. You interact it with it through scores and reports. There’s not a chat box or a chat window where you can ask things, right? It’s a pre-built tool or a tool that’s built for you that runs behind the scenes. Data is fed to it through a system, and data is exported from it back into a system, so you have access to these scores and reports. Fundraising example: these 50 donors are most likely to upgrade.

Speaker 2 45:49
It’s taking a lot of data and telling you what to do with someone, when, how they’re going to respond in the future, so that you know where you should be piecing them into your portfolio and into your, your process conversations that you’re going to be having, the function is it prioritizes prioritizes the right people way better than we as humans could by identifying those underlying indicators that we wouldn’t be able to catch ourselves or flag ourselves. It’s doing this work for us in a way more efficiently in a week of it. Current adoption minimal 2.3% It addresses the donors down dollars down by putting the right people in front of us at the right time. Generative creates content, synthesizes a lot of data, creates packages it in a way based on how you’re asking for it, what you’re looking for, and creates new content in that sense. You interacted with prompts, with by editing outputs. It’s very, it’s much more reactive, much more dynamic, much more interactive. The typical chat window that you see with ChatGPT, there’s voice modes, it can do coding, it could do all these different things, but it is much more responsive to than what we see with predictive AI, based on how it’s designed, fundraising example. Here’s a draft appeal in your voice for spring, so it’s creating that content for you. The function is to personalize information, either for you or for the person that you’re sending it out to, but really you and the 400 people attending this, like, we all learn differently, we all think differently, we all speak and feel and perceive things differently. So, how do I.. we take the same information and personalize it in a way that meets us where we’re at. I love tables, which is why we’re all looking at the table right now. But this might not work for everybody. A tool that I’m not going to recommend, many tools here, but one thing, just kind of like off the cuff, Notebook LM, Google dot notebook lm.com I think, something like that, but notebook lm, it is a great tool for synthesizing information in different ways. You give it a website, you tell it to research something, you give it a pdf, the pope’s 42,000 word encyclical, and it can create a podcast for you, can create an infographic for you, can create a video for you, like slides for you, all these different things. So, no matter how you like to learn things, that is a tool that is very tailored to each individual, and you can make it specific for your own, so that you can gather more information more readily. That was a quick aside. Current adoption, 86% are exploring, so much deeper adoption in this than what we’re seeing with predictive AI, and it address overworked and under-resourced staff. One caveat here: generative AI can be very convincing. It can tell you a lot of things, and it can sound incredibly right, even if it’s not right. Right, there are hallucinations are a thing, but I’m sure in my life I’ve sounded very convincing, and for something that I was found out later to the entry, right. Regardless, my concern, and it’s probably because I’m so deep in the predictive AI space, and that’s kind of where my history is, is that organizations will find predictive AI to be a valuable tool that they should be leveraging, leveraging, but then go to generative AI and say, hey, can you predict who’s going to do what, and when AI is generative AI, is going to sound convincing that it is doing that, but it is not actually doing that right. Generative AI is not building a predictive model, maybe in the future it can, but right now it cannot build a predictive model to say who’s going to do what and when. So do not use it for that, because if you do, you’re going to get the same results as if you used any other methods, and then you’re going to be dismayed by predictive AI when it wasn’t really predictive AI in the first place. That’s kind of what keeps me up at night. There’s actually a commercial by Microsoft, where I don’t know if you saw it, it’s like the NFL Combine, and they said predict which running back is going to have 600 yards.

Speaker 2 49:39
I don’t watch football, so I could be wildly off of that, but it’s using Copilot to predict when it’s actually not predicting, so like there there is concern that people are going to use the tools the wrong way, which is why we’re having this conversation to help demystify and help determine where these tools are most applicable. Finally, Agent. This one executes workflows. It does not just create content, does not just do one task. It executes a workflow. When we think about our work, we’re executing workflows in many times a day, right? So it helps move through some of those steps a little bit quicker by gathering a little bit more information, and so on. You interact with it by telling it what you want to do, setting a goal, and then approving checkpoints along the way. A lot of times, these built-in systems will make sure that they confirm with you before taking any specific action. They have certain, like, thresholds of risk, and so they’ll say, “Okay, I’m going to reach out to you when something is over this threshold, but if it’s under this threshold, I’m not going to reach out to you, and I’m just going to act autonomously. That threshold, I think, can be changed, but the old tools are different, and they all change all the time. So, like, there is opportunities, and it requests human interaction in certain steps to make sure that it’s not acting wildly. But things have gone off the rails. You can see on X or Twitter someone who is in security at, I think OpenAI, they were using a tool called Open Claw, and it deleted, like, hard deleted 50,000 of their emails, even when it was telling us not, so it’s still like it’s still learning, right? But it is getting better every day. Fundraising example: I’ve surfaced donors, cute outreach, and tracked every touchpoint, that is something that agentic workflow could do, right? You point it to what you want it to be prioritizing. If you say identify people based on capacity versus predictive, like you know what results you’re going to get, right? But it needs to know exactly what types of tasks it’s supposed to take, and then what the end goal is, and then it can execute on your behalf. Function is obviously executes 1.2% are using this crazy that predictive AI, which has been around for a decade, only has twice the adoption rate of generic of agentic AI. When agentic AI has been around for effectively like seven months, it addresses budget constraints and staffing shortages, so really, like this is key. This is a do a lot of work with nonprofits. I’m using AI, I talk about AI a lot, I think about AI a lot, and I think about the nonprofit sector a lot. Like, my, I’ve been in this nonprofit sector for half of my life, my entire career, and I’m committed to the work that we do. We do incredible work, and it’s an honor to work alongside all of you in this right. However, using AI, even though it’s accessible and it’s available, and it could do all these things, it’s not always used the right way, right? A lot of organizations are not using it the right way. A lot of organizations are buying tools, not using it right, but also a lot of organizations are using it where it shouldn’t be applied. I think, again, like you probably heard me. We have seven minutes left, is that right? Oh, and you’re, you’re muted. Sorry. Anyway, the culture is up to you. I’ll wrap up quickly here. Culture is up to you. The main thing about AI adoption is leaning into what we can do as humans and making sure that we’re keeping AI separate from what we should be prioritizing, which is those human relationships. A lot of nonprofits are using AI, but only 7% are identifying it as information as impactful in their transformative in their work. These numbers at the bottom are repeated over and over again. We’ve seen that in my time in AI adoption that 70% of successful AI adoption is based on the culture itself, right? A couple resources I’m going to go through as quickly. Fundraising AI, it’s a membership organization. It’s free for everybody. We use a responsible, beneficial framework for AI adoption for nonprofits, knowing that we’re different.

Speaker 2 53:50
So, if you haven’t gone there before, please take a look at it. Your steps.. I think these slides are going to be available again. Apologies, I get passionate and get caught up in stuff, but I do want to take some time for questions, if you want to download these slides, you can use this QR code. Okay, questions.. let’s see, what do we have here? Can we use AI tools within DonorPerfect to do these things, or is it automated outside of DP? That is more of a donor perfect question. Unfortunately, I’m not able to answer that at this time, but your DP rep should be able to address that for you. Next question: What are predictive AI tools that are both safe and ethical to use and protect donor information, but don’t require entire product bottles, are free or affordable for small and midsize nonprofits. Protecting donor information is key, right? Like, we using your data to put it out in the public is not what we want to do. So, enhanced core models that are available through DonorPerfect, or enhanced core models through DonorSearch, but you’re using DonorPerfect. Have access to enhanced core models directly. Those are built for nonprofit fundraising sector. We are not publicly making your data available or keeping it private for your organization. How do we ensure that the gender-related tools don’t just continue to raise expectations placed on fundraisers and marketers to increase productivity and goals and continue burnout. How do we ensure that the generative AI tools don’t just continue to raise those patients? Ah, I see by offloading the time and giving them some of their time back, that they just have to do more with that time than they available, and then continue the burnout. That’s a very good question, that is, that’s the leadership, that’s the leadership of the organization, being able to recognize that the same work is getting done, that the goals are being met, and so, how do we use that time effectively without just going deeper and doubling down on burnout? That’s a big thing that’s coming up. There’s a lot of different articles around that, but that is a human element of leadership and prioritizing their team as individuals, and definitely a good thing to raise at the organization you’re at. How do we rectify the environmental and community damage caused by edit data centers in direct contrast to our mission? That question comes up many, almost every time that I talk about this, the it, there are significant impacts that it’s having, right? I do think that there are impacts that are being misrepresented, negative impacts that are being misrepresented, but that doesn’t mean that there is no negative impact, right. Generative AI is takes a lot of energy when new models are being built, right. Using the model does not take nearly as much energy. Back a year ago, a report came out that saying that a prompt, an AI prompt, is the equivalent energy usage of seven seconds of streaming Netflix. Right, we all stream Netflix. I watched a movie last night that, in the end, I didn’t even really care to watch, but I just did it because out of boredom, right? So, like, we all do these things, we make these choices, but what’s working in our favor is that these AI companies, these AI labs, are just as committed to identifying energy-efficient ways to provide this product, albeit for different reasons, less concerned about the environment, probably more concerned about profit, and the more money they spend, less profit they make. So they’re trying to figure out how to be as energy efficient as possible. A lot of data centers are moving to space because it’s constantly cooled, there’s no, there’s less taxes up there, and so like we’re seeing innovative ways, and also predictive AI can help us identify ways that can be more energy efficient that we wouldn’t have been able to identify on our own. So, these are all all things that we can expect to see as we continue on. I think that’s it for me. I really appreciate everybody attending. I hope you enjoy the rest of your conference, and please feel free to reach out to me if you have any questions or want to continue this conversation, it’s been really, really wonderful having this time with you, so thank you all, thank

Speaker 3 58:22
DonorPerfect

Speaker 4 58:42
is what we consider a fundraising CRM, so it’s very purpose-built for fundraising. Donorsearch has been our go-to as far as the standard for data pen services. Their data has always been top notch. They’ve communicated to me what their value proposition is, in terms of what differentiates them, and that has just made the difference in the world, as far as being a good partner to us.

Speaker 5 59:10
That’s where our partners come in. It really allows us to expand our footprint and the amount of people that we can help to, I mean, millions of people all across the country,

Speaker 6 59:22
we really pride ourselves in ensuring that our clients not only meet their goals when it comes to fundraising, whether it’s capital campaigns or direct mail, whatever it might be, but exceeding those goals.

Speaker 4 59:34
Integration allows the data to talk and merge into basically a single database, so that when you’re looking at a donor perfect system, you’re looking at everything that you know about that constituent, that donor, that volunteer, whoever they are. I see the value of being able to tie two systems together in terms of the sheer hours that are saved on a day-to-day basis. As well as really having a five quality data,

Speaker 5 1:00:03
our tools make it so much easier for them to fundraise more effectively.

Speaker 4 1:00:08
For many, many years, our clients have relied on that partnership to be able to bring in wealth, so that they can screen their donors and figure out who they should be communicating with, as far as high capacity and really a good return on the investment.

Speaker 5 1:00:21
It really feels like we are making a difference on a day-to-day basis. It’s probably one of the greatest thrills of the job is actually getting to work with these people, and and bringing the new solution for them that changes how they’re doing things. It’s changing what their best practices are. We really help you get in there and bring the data where you need it most

Unknown Speaker 1:00:41
and.