Transcript
Okie dokie. It is noon. Let's get started right on time. Hello everyone and welcome to our webinar today from overwhelm to optimize which will help nonprofits learn to use AI to raise more without getting overwhelmed. My name is Julie Kennon. I'm the director of marketing at CharityEngine and I'm joined by my boss Dave Martin, the vice president of marketing at CharityEngine. Um, from the top, from the top,
I'll say we're not an AI company. We're not pretending to be an AI company, but we are learning how to use AI to work smarter on our own team, and we're helping a lot of our nonprofit clients do the same. So, this webinar is created for the person that hears, "You need to quote unquote use AI, but you're not given a roadmap." That will change in the next 30 minutes. I promise. No hype. A very practical path. You'll learn prompts you can use. You'll learn governance. You'll learn all the basics you need to get started with AI. This is not a technical deep dive. This is a very high-level intro course. We are happy to do another webinar if you want to connect AI to your donor data and do things like that that are a little bit more sophisticated, but this is just sort of a beginner presentation.
So, in the next 30 minutes, what will you learn? What needs to be in place before AI will help you? How to write prompts? and we'll show you prompts and we'll do this live. How to write prompts that really work. There is one rule that keeps donor data safe. This is the governance issue. We'll offer a simple 90-day plan that you can start today or tomorrow if you really want to jump into AI. And we'll give you a live demo and a playbook that Niha will put in the chat with worksheets that can help.
Now, this is a quick little special offer. I think Dave will mention this again at the end. This is for webinar attendees only. When we offer this in our webinars, uh, people really love it. It's a data health score that you get. It's a quick 15-minute call. We'll look at your systems, tell you where automation gaps are slowing you down, and tell you where quick AI wins could help you improve fundraising. Now, it's the fastest way to find out where AI can help your organization specifically.
Okay, let's take the pressure off. If you are feeling behind on AI, you are not. I promise you, everyone in this space is figuring it out in real time, including us, including the big tech companies. And starting today is going to put you ahead of most of your peers. No one's mastered AI yet because it is evolving too quickly. The goal is not mastery. The goal is reclaiming time. If you've been waiting to get it right before you get started, don't wait. The time to start is right now. It's moving fast and you're not behind. Nobody has it all figured out.
So, when we're talking about reclaiming time, what does that look like? If everyone on your team could save three hours a week automating things with AI, using AI to help with emails, subject lines, etc. That's 150 hours a year per person back to your donors and back to your mission. If you scale that across a team of five, that's 750 hours, which is basically a part-time employee at no cost to you. And that's the real AI ROI. It's not flashy tech. It's just getting hours back. And everything we're covering today is going to help you do that.
So, here's the question. Where do you start? Right before we answer it, we want to talk about why it matters. Because donor expectations have already shifted. Your donors are being shaped by AI, even if they don't know it yet. They expect personalization. Everything they see today is shaped by personalized AI. Whether that's Amazon, DoorDash, Uber, Netflix, when your communications feel generic, it'll register. They might not be able to name why, but they're very used to, oh, Julie, you liked watching Succession. I think you would like this. I'm used to seeing that. And if I get an email that doesn't call me by name, doesn't talk about something personal, it just kind of feels weird. And your donors are not comparing you to other nonprofits. They're comparing you to every digital experience they've ever had.
So you're being compared to all of these. When you are slow, when your signals are slow, your website is slow, your emails are slow, they're very generic. It signals you don't know me, you don't know anything about me. When everything is fast, when your responses are very quick, when they're automated, when they're personalized, you feel trustworthy. So, a slow generic thank you isn't just forgettable. It's going to erode trust. They want to feel like you are on it. You know them and you are responding to them every time they reach out to you.
Now, this is interesting. Half of all nonprofits use AI in some way or trying to use AI in some way, but three-quarters have no strategy behind it. The gap is the opportunity where you can jump in and start using it. If you walk out of this webinar with a plan, you are already ahead of most of your peers and that's why we're here.
Now, a little jump ahead. Eventually, you're going to want to use AI on your donor data. You're going to want to identify lapsed donors. You're going to want to identify major gift prospects before you can find them. And there are a couple things that have to be in place before AI can work effectively for you. So, as you're getting started with AI, I want to show you this one important point. AI will amplify what you already have. If your foundation is shaky, it's going to magnify the chaos. If you have disconnected data, if your info lives in three places, AI is going to send three emails to the same person. It's just going to do it faster. If your donor journeys are unclear, your $25 first-time donor is going to get the same email as your $5,000 major gift prospect, only faster and incomplete records. Dear John could go to somebody whose name is Jennifer. AI is not going to fix your dirty data. It's just going to scale the problem. So, as you grow into AI, you'll want to identify things like lapsed donors, as I mentioned, and you have to clean your data first. So just over the next couple of months, start to think about how you can clean your data and get it in order for AI to use it effectively.
Okay, so this is where most nonprofits get stuck. You know, the foundation matters. It feels way too big to fix. And I'm here to tell you, you don't have to fix everything before you start. Start really small, see results, and build momentum. So I'm going to show you four things you can do this week that are really easy.
One, rewrite emails. Paste a hundred-word thank you email that you've sent into ChatGPT or Claude or any of these programs. Ask it to be warmer and more specific and then compare the two. That'll take you 30 seconds. Don't agonize over which subject line is going to be perfect for an email. Ask AI to generate five of them and pick the best one. It'll take you less than a minute. Build a welcome sequence. Do a 30-day new donor series with a single prompt. Now, this will give you the structure. You'll obviously want to go in and edit it and make it more personal, but you'll get the structure very quickly. And optimize your donation page. Take your donation page exactly as it is. Put it in ChatGPT and say, "Where's the friction?" This is not new software. This is not using your donor data. This is just using what you've already got and optimizing it with AI.
Now, let's look at what this looks like. "Thank you for your support. You're making a difference." It's nine words, generic, skimmed in 3 seconds. It's very forgettable. Every nonprofit sends an email like this. It's some version of a generic email. And that's the problem.
After you ask AI to take a look at it, look at this. It's specific. It's harder to ignore. We name the program. It's Full Plates Friday. We mention the exact gift they gave us. It's $75. We create an image in their head. There's a student going home with a bag of meals because of you. We showed the scale of what we do. 1,200 kids every week. And then at the end, we dropped in a small monthly ask that's very gentle. If you give AI those specifics, and this is exactly what Dave's going to show you in about five minutes — if you give AI specific information, it will give you the results that you want. It'll just be a little bit more polished and more effective. And this is what's changed. The second one was specific, personal, and immediate. That's what AI does when you give it the right inputs.
Now, why do most prompts fail? Well, if you say, "Write a nice donor email," AI is saying, "Well, what's the tone you want? What are you asking them for? What donors are you talking to?" That's really vague. It's just going to produce kind of generic garbage. If you add structure, you're going to get better output. AI will use everything that you give it, and where there's a blank, it's just going to guess. So when you add structure, you're going to get better output. Specific inputs like your audience, your tone, your constraints — like don't use the words "giving opportunity" — will drive more specific results.
And here are two tricks that will dramatically improve what AI gives you. Start broad and then refine. Don't ever think the first output AI gives you is going to be perfect. It won't be. And that's okay. You can iterate. You can say make it warmer, make it shorter, make it funnier, make it less corporate. And treat it like a conversation rather than a vending machine. Keep telling AI what you want, and it will keep producing something until it gets to exactly what you want to see.
Another good idea is to show it what good looks like rather than telling it. Take a great email that you've written, paste it into ChatGPT, and say, "Match this tone. It'll sound exactly like you." And tell AI again what not to do. Don't use exclamation points. Don't use em dashes. Don't use the word "actually" because it loves the word "actually." And just give it examples until it matches what you want.
Now, this is a big question. Is donor data safe when you're using AI? If you open ChatGPT, is all of your donor data going to flow into it? Here's the clear answer. These are the things that you do not put into AI: donor names, email addresses, giving amounts tied to a real person, personal notes from your CRM like "Bob's birthday is Tuesday," payment or credit card data. And this is the number one rule that we talked about in the beginning. If you wouldn't be comfortable putting something on your website, do not put it into AI.
What's fine to use? Mission and program descriptions. Anything that's publicly available, totally fine to use. Draft copy that you want to improve, anonymous stats like "we serve 1,200 kids weekly," and meeting notes with names removed.
One note — Dave was asking me about this the other day and we were talking about it. If you use a paid Claude account, does that mean that your data is safer than a free account? And the answer in Claude at least is no. Claude trains itself on chats. If you have a paid team account, it does not train itself on chats. So, what you could do as a nonprofit team is create a paid Claude team account and then it would be safer. Um, that's not legal talk. I'm just telling you that's how we operate and that is what we have learned.
Yeah. And what I would just add there, Julie, is that if you get a paid Claude account, just an individual one, that's what Julie's saying is not going to necessarily keep your data within the walls. It's when you get like a business type of account. We have a team account where we have five users in our team account and then there's like enterprise level for really big organizations. But both of those more business accounts are going to keep your data from being used for Claude training itself and able to give answers to others, right? So, that's what we learned this week in particular and that's one of the reasons why we have a team account. It's a pretty important distinction and a question that comes up often.
Now, here's a mistake we see a lot. People get really excited about AI. They run back to the office and they want to AI absolutely everything. Emails, reactivation, events, grants, social media, everything. And in three weeks, it completely falls apart. If you are juggling everything, you're probably going to drop everything. So, our advice is to pick one path and do it end to end, get the results. A new donor welcome is a really easy first pick. It's self-contained. There's a clear goal. You get your results quickly. An upgrade to monthly giving is another good one. And a lapsed donor re-engagement is another one. Start with one, finish it, try the next one. What this does, it gives you a quick win. You use it, you realize it works. The second one's a little bit easier, a little bit faster, and your board is like, "Whoa, you guys are rocking it with this AI. Keep it up." That helps, too.
Then this is how you go from trying AI to using AI. A lot of nonprofits get stuck in trying. They dabble. They experiment. They never really move forward. Here's how you actually use it with a plan.
Month one: dedup your CRM records, at least the past two years. Write a one-page internal AI policy — what's okay, what's not. Get two or three core prompts and train your staff on them. Go into the different ChatGPT, AI, Claude, Perplexity, all the different free engines. Put that prompt in and see what it's spitting back out at you. When you really find the exact prompt that works, keep that and give it to your team. Then identify one donor journey that you want to rebuild.
In month two, build that donor journey. Have two different subject lines that ChatGPT or Claude gives you and send those emails to different audiences. See which one gets more donations, more conversions. I'm such a big fan of this. Use AI to audit your donation form. It will quickly tell you you have too many fields, you need a monthly giving option, get rid of your video — whatever it is. It will give you good feedback and you guys can change that within an hour. You could optimize your donation form.
And then finally, scale. Add your second donor journey. Now it's easier. Launch a monthly giving upgrade sequence. Recurring revenue, of course, we know is the gift that keeps giving. Enroll your ChatGPT, get them to help you, and build a simple dashboard. You want to show results. It's really important. This is a 90-day plan. No consultant, no bigger budget. Just get started.
Okay. Now, Dave, I think you're up.
Awesome. All right. Good. That was good timing, Julie. Um, and great setting the stage here. I'm just reminding you guys this cool offer that we do — it's a free fundraising data health check. We'll meet with you for like 15-20 minutes, ask you some questions and give you a really structured output of how you can improve your data health. And that's something that's critical obviously when you're working with AI anyway. So anyway, five spots available.
All right. So we're going to do three levels of AI prompting here for nonprofit fundraising email campaigns. This is just an example and this is going to be wild and live. So we'll see what happens. I can't promise what the outputs are going to be. We're going to do level one which is a beginner prompt — it's like a just-get-me-started type of prompt. And then there's level two, intermediate — you're giving more inputs and getting more control over the output. And then the advanced one, which is like you're making it as good as a professional copywriter would make the email series. So you've got three different levels. I'm going to go through each one of these prompts and show you guys. We're going to share the prompts at the end. You can take them and use them. And in particular the advanced prompt has got some structure to it that you would need to edit for your own organization. But I'm telling you it's really cool.
All three of these prompts have the same goal — they're all writing a three-part fundraising email series. So that's what we're instructing the AI to do. And this is the key thing that we want you guys to get out of this: quality. The quality of what comes out reflects the quality of what goes in. So the more specific you are and the more structure and the more guidance you give to the AI, the better quality you're going to get out and the less time you'll need to do any editing. That's really critical.
All right. So let me share my screen. Let me pull up my prompts and I'm going to share my window. Okay, cool. So I blew it up. I'm hoping you guys can see this.
So again, three prompt levels, same goal. Everybody leaves with a prompt and garbage in, garbage out. Let's start with — oh, and I would point out one thing real quick: Dave is not logged in. These AI tools will remember things that you tell them so you can reference previous conversations if you're logged in. He is not logged in. This is exactly what you would get if you just Googled ChatGPT and went to it.
Yeah. And as we were doing dry runs going through this, that's what we learned. It's better to be not logged in. And by the way, we used AI to do a real lot of this webinar — whether it was the illustrations that you saw, the structure of it. We're using AI throughout this.
So, let's start with the beginner prompt. Here's your first prompt: "Write a three-part email fundraising series for Bright Path Literacy" — this is a fictitious organization — "a nonprofit that provides after-school reading tutors and book libraries for kids in rural Appalachia. The goal is to raise $90,000 by December 31st."
You can swap out Bright Path for your own organization and description and what your goal might be. But I'm kind of curious — what do you guys think the output's going to look like? What will the AI have to guess?
Elizabeth says format of output. Odette says audience — 100%, that's right. The tone — there you go, Kristen. Not your nonprofit's status necessarily, but the audience, the tone, what the story might be, the urgency. Let's see what the output looks like here.
Okay. So, it's giving us three parts. We're not going to go through all three emails. We're just going to compare the first email for each one of the prompts.
Subject: "A child is waiting for a book tonight." Like, okay, good. It's got my attention. "Right now in a small town tucked into the hills of Appalachia, a child is finishing their school day without a single book to take home. No stories to get lost in, no extra help sounding out words, at Bright Path Legacy..." Right? Like they do state what their goal is further down in the email. It's telling a little bit about a kid that is not able to take a single book home, a little bit about the organization, and then what the goal is. "We can't do it alone," and they're asking what the gift is going to give. Like, this is legit. I could use this. It's competent. It's a decent email.
Any thoughts on what this feels like to you guys? For me, there isn't really a sense of urgency. It's not very specific to Bright Path itself. It's competent, right? But it's generic. Kristen says it feels like AI. You're gonna have to edit this series, you know. You're probably going to spend 30 to 45 minutes editing this.
Julie, any thoughts? Yeah, I think one thing that you guys are bringing up is that as more and more people are using ChatGPT and AI, there are those tells — you have the em dash in here. You can start to tell when things sound like AI wrote them. And that's why Dave's saying it's going to take 30 to 45 minutes. Take the bones of this email and you'd want to start to rewrite it and put it in your own voice, the voice of your nonprofit. But no, this is generic. It's got a little bit of information. It's really guessing the rest and kind of putting in a lot of filler story type material. So, let's go to level two and see how it changes.
And Amy, yes, email A/B testing is always a good idea. You change one thing — the subject line, the ask, how much money you're asking for, how much detail you give. You change one thing and you test two and you see which one converts better and that's how you dial in the perfect appeal.
So the output is competent. It just might not sound like your organization. And you've got to edit it. So let's see what happens now when we give AI something real to work with.
We're going to go to prompt number two, which is the intermediate prompt. So right out of the gate: "You're an experienced nonprofit fundraising copywriter." So it starts with the role. It's telling the AI how to think, who it is and what it's doing — it's writing this three-part email series.
And then it gives more context. So number one is the role. Two is more about the organization. So yes, Bright Path Literacy, it does after-school stuff and take-home book libraries for K through 5 students in rural Appalachia. Getting a little bit more detail here for the organization and what it does and who it serves. The goal is raise $90,000 by December 31st to fund 500 student reading kits. Okay, we're giving it more input here, more specifics.
The key one here is audience. We're telling the AI who it's writing to. Your audience is mid-level donors who have given between $100 to have supported Bright Path for two or more years. It's very specific. And just this line alone changes everything about the copy and what's going to come out. We also have a key message: every $180 funds a full reading kit for one child for an entire school year. We also tell the tone — warm, personal, urgent, written as if it's from the executive director, not the organization. We don't want it to sound like a press release.
And then there's a structural ask: each email should have a subject line, preview text, and a clear call to action. Space the emails one week apart, and write the final email going out December 29th. Okay, so we had role, organization type, the goal, the audience, the key message, and the tone instruction. Let's see what happens.
As it's writing here — let me answer Marissa real quick. Marissa said, "You'll want to reference how they entered your email list. Did they opt in for an event, a free resource, etc." And Marissa, this is the next step. This is where you do want to use AI on your donor data. You can actually connect Claude to your CRM. You can do different things. And then you would say go to the records, figure out how they entered the database and reference that in the email. So it gets really cool. You're right, that would personalize it very very easily.
Now let's look at what happens here with us giving the six different inputs — this intermediate prompt.
"A child is waiting for their first full reading kit." Okay, you got my attention. "As I sit down to write you, I keep thinking about one of our students, a second grader who told her tutor, 'I didn't know books could be mine.' That moment has stayed with me." So it's starting to tell a story here. "Because of the many children we serve across rural Appalachia, books aren't a given. Reading support isn't a given. Opportunity isn't a given. But you've been helping change that." So recognizing — here you go — we already said they've given before. "Thanks to your past support, Bright Path has been able to provide after-school tutoring and take-home book libraries to students who need it most." It's pulling in some of those details from the key message — how each kit costs $180 and provides a child with a full year of after-school tutoring and a collection of books. And there's some urgency — they've got to raise $90,000 by December 31st to fund 500 reading kits.
What do you guys see in this email versus the first one? The first one was a little bit more generic-sounding now. Even though it was good before, it's sounding more generic versus "As I sit down to write you, I keep..." It's personal. There's more context. The story is more robust. The ask has more context. What are your dollars going towards? What are you helping this child? I like the first-person narrative.
There's urgency in this. There's specifics. I think I would spend less time editing. Marissa's exactly right. Giving it the role, and it's coming from the first-person narrative, from the perspective of an executive director. ChatGPT kind of nailed that. It's got the data points we gave it — the $180, the $90,000, the 500 reading kits.
The only thing that I would do if I were writing this email — the next prompt would be "make it shorter." And that's what we're saying you would do. You would go back and you would refine each of these. So it's not necessarily editing it with a pen. It's refining it through the tool.
It's interesting — it did give us the preview text, but it gave it at the bottom, which is kind of funny. Other times when we've done this, Julie, it gave it right below the subject up top. And sometimes ChatGPT will give you three versions of a subject line to go through. It's different every time. So you might need to do that back and forth to hone it, but you can easily just copy and paste it.
Odette says definitely an improvement. As a donor she would be interested, but the fact that the story in the start is fictitious bothers her and it feels made up. Odette, you could put in a story from your perspective and say, "Hey, we have a kid that said XYZ." And then it's not made up. You could put that as part of the prompt. Exactly.
So all right, let's look at level three. The edits become meaningful here. They're not corrective — and that was one of the key points we wanted to make.
Let me go pull prompt number three. It's a little bit longer. Dave always tells me not to talk so fast, and I always tell him to move it along.
Okay. So let's look at this prompt. This one is long. This is going to look scary, but it's not. And you'll see how you could put your own details in it.
There's a prompt summary of what it does and when to use it. The role here: "You're a seasoned nonprofit fundraising strategist and an expert copywriter with deep expertise in donor psychology, email campaign architecture, direct response writing, and cause-based storytelling. You have helped organizations raise millions of dollars through compelling multi-touch email campaigns." The role here now is much more in depth.
But the action plan is the key thing. The action plan tells the AI to go and learn best practices before it writes. So learn, then teach itself. The prompt teaches itself before it writes, which is kind of mind-blowing. "Share your comprehensive knowledge of best practices for writing high-performing nonprofit fundraising email series. Specifically, what makes loyal mid-level donors open, read, and give again. The ideal structure for the three-part series. Flag common mistakes to avoid, such as leading with organizational history instead of a child's story or using guilt-based language that erodes the trust these donors already have with Bright Path." Then "apply these best practices to write all three emails using the campaign details below." Wow, that's kind of cool. You're adding in the action plan — telling the AI to learn first.
The context here is really guiding the AI to think from the donor's perspective. The audience is mid-level donors, $100 to $500, who have given for two years or more and have a real relationship with this cause. "Don't assume they need to be convinced of the mission. They already believe in it. Focus on making them feel the urgency of this specific moment and the tangible impact of their gift. Build emotional intensity across three emails as December 31st approaches."
And guys, a quick trick you could use on any of these: you can always say, "Ask me questions until you have enough information to give me the output I'm looking for." So you can literally say to ChatGPT, "What else do you need to know to give me a really good answer?" That's a little trick that works great.
Part of the execution here: we're telling the output to be three complete emails each with a subject line, preview text, salutation, body copy, call to action, and a suggested send date. And it also says "success is measured by a series a fundraiser could send tomorrow with minimal edits." So you're telling it what success looks like.
"Ask me questions until you have enough information to successfully complete the task." We're going to see if it does that because it's hit or miss on this one.
All right. It's asking questions. "What's the ED's name and tone?" The ED's name is Julie Kennan. And her tone is warm and friendly. And she uses first-person singular. Then just get it to rerun it with that one thing. Let's see what we get.
The emotional connection, story-driven opening, suggested send date — December 27th, it was year-end. So it's giving us some optional subject lines and preview text up at the top. "Hi [first name]." Late Tuesday afternoon, I sat at a small table with a second grader named Ava. She was holding a book in her lap, very still, almost like she was afraid to move. Her tutor gently encouraged her, "Go ahead. You've got this." Ava looked down at the page and whispered the first word, then the next, and then something shifted. Like it's telling this story and you're already getting pulled into it. "And moments like this are why Bright Path Literacy exists and they're only possible because of people like you. Right now we're working to place 500 reading kits." They've got a sense of urgency in there. "Each kit costs" — connecting it to Ava and what the outcome's going to be, where your money is going towards. "Would you consider funding a reading kit today? Your gift of $180 ensures one child gets a full year of support. And if that's not the right gift, any gift moves us closer to that goal. — Julie Kennan, Executive Director."
And then you want to show the first email again just to contrast. First one: "Right now in a small town tucked into the hills of Appalachia, a child is finishing..." There's no name. There's no story. It wasn't bad the first one. But now you're getting much more specific. There's a different tone — it's your tone, warm and friendly, telling a story. You've got the action plan of learning best practices before it went into it. You've got "funding a reading kit" as a very clear CTA at the end.
It's like it blows my mind. And by the way, I took a prompt from somebody that I follow in the marketing field. I took one of his prompts and I told Claude to make this prompt for this webinar and made the three prompts based on his. So you can get all kinds of stuff, but following industry leaders on this stuff is really useful and gives you good ideas.
The lead was the child's story — not about the organization's history. Best practice. It's the emotional hook for email number one. There's no guilt-based language and there's urgency. This is really exciting.
Julie, any other thoughts on the three different prompts here and the outputs from them? No, I think you really clearly demonstrated the better information you give it, the better output you're going to get.
If you stop sharing, we can go to the elements — tips for getting the most from any prompt — and then guys, we're just about done.
Tips for getting the most: start with level one. Start small and build on it. Julie talked about that earlier. The iterative approach is where you're going to get the most out of it. You can iterate and give more detail — get one version, say "okay, let's tweak it a little bit more here," and you can continue to give it more. But if you have a prompt that's very specific, you're going to be much closer to what you need in the end. That last one — it was maybe a few minutes of edits. It's not a lot compared to the first one. But start small and build on it.
Let the AI ask you questions. If you have that level three prompt, having it say "ask me questions until you have enough information" — we could have continued to answer those questions and the output would have been better. We just didn't have time for that.
What's next: swap the channel. What you can do here is replace the three-part email series with a direct mail appeal or a social media campaign you're going to do. And the rest of that prompt is going to hold up. So you can take this prompt and create it for other channels. You can even tell the prompt you want it to create a landing page, the email series, and the social media post to go along with it, including the images. There's a lot you can work with the prompts.
Garbage in, garbage out. Quality in, quality out. The more guidance you give, the less guessing the AI is doing and the better output you're going to get, the less editing you're going to have to do. But obviously you want to make it sound like it's from your ED or whoever it is from your organization. The more details you give it, the better.
We've started one of the projects within our instance of Claude — you have these projects. I just created it today, which is an AI documentation library. So I'm putting a lot of background information in there like about our buyer persona. And you guys might have a donor persona and a donor journey. The more information you can put in there that Claude can reference and use and learn from, the better it's going to be.
Golden rules: quality in, quality out. Any questions?
I'll tell you a quick funny story. My kids make fun of me for using ChatGPT for things and we went hiking over the weekend and I took a picture of a trail marker and I said, "Which way do I go to find the blue trail?" And ChatGPT got so sassy and it wrote back, "Julie, that literally says the blue trail is here." And I was like, "Well, okay. All righty then." The kids were like, "See, even ChatGPT thinks you're annoying." It was very funny.
Yeah. My feeling in all this is just embrace it, don't be afraid. Test it. Try things out. Take a course, learn from people that have been doing this for a longer period of time. Think about the prompts and the level of information that you can put into a prompt. Ask ChatGPT or Claude — when you've got a paid account and you've got it firewalled — to create a prompt for you based on your organization. You could take our prompt level three and say "take this as an example to create one for our organization" and you'd be surprised what you get. Iterate, go back and forth through it a little bit, and then use that. You can create prompts — I mean we're looking at using prompts for data management right now and creating agents for data management. Go into HubSpot, pull these numbers that we want to report on every day, analyze them, and tell us the trends and spit it out, and then we'll talk about it as a team.
And if you've got data silos and your data is all over the place, you're going to have to be uploading Excel spreadsheets into Claude or ChatGPT and it's going to get messier. And as you get more proficient with AI, you're going to want to connect it to your CRM. We connect to HubSpot. There's a protocol that Anthropic — which owns Claude — has built specifically for connecting HubSpot. And so I can query using Claude into our HubSpot database just like you guys would be able to go into the database to get information about aggregate data, anonymized data about your donors, and maybe trends that you're seeing for a board report you have to do. There's just so many use cases. It's endless. But just get started. Don't be afraid of it. Get in there. Use some of these prompts. I hope they're helpful.
Two quick things: Nha shared this AI playbook. This is a very general guide that we wrote. It's got a list of 20 prompts in it. It's got all sorts of tools. Download that. It's very helpful and gives you a little bit of a blueprint and dovetails with this presentation.
And Amy, you asked how we're using AI as a team. The same way that we would say to you — take your most routine tasks and get AI to help you. We're building agents to do things. Go into HubSpot, pull these numbers that we want to report on every day, analyze them, and tell us the trends and spit it out, and then we'll talk about it as a team.
Odette, thank you. I am an old dog learning new tricks, too. And at first I thought AI was just like witchcraft. And then all of a sudden, I realized it's incredible. It will make your life so much easier. And it will help you raise more money. It will help you engage your donors better. And I can promise you that's true.
If you want that data health check, do that because it's a really quick way to figure out how you can use AI immediately. Odette — grant applications — human input and expertise is critical, but AI has helped you optimize those applications. Perfect use case.
What platforms are we recommending? We like our own platform. And I would say too, Marissa, it really depends on the organization. The fair answer is it depends on the organization, how big they are, what your needs are, what your challenges are. If you're just starting out or if you've been doing this and you're a $20 million a year organization, then the platforms are very different. We tend to work with organizations anywhere from like $500,000 a year all the way up to $100 million a year organizations. Wounded Warrior Project is one of our bigger customers and we've been working with them for years. Really depends a lot. There are some platforms that are great for just getting started. And then there are other platforms like ours where you have all the tools you need in one place, so your data is in one place and you're not having data silos and all that importing and exporting.
That's not really a plug for us. The answer is it depends. But we're proud of the work that we do. Especially when it comes to sustainer programs — if you have a monthly giving program, our platform is the best in the industry when it comes to retaining that revenue from your monthly donors, and there's AI involved in that a lot — features that make us really an industry leader there. So, it depends. If you had a lot of monthly donors, we'd say "you've got to work with us."
If there are no other questions, we will wrap it up. You've got our email addresses if you have feedback or you want to talk more. Thank you for coming today. I hope you learned a little bit and we will see you again. Bye guys. Bye.
