AI Fundraising Build Pt. 4: AI with Catholic Guardrails

 

Episode 11, October 2, 2026

 

 

Show Notes:

What is Pete AI, and why should Catholic fundraisers trust it to support their work?

 

In this discussion episode, we explain how Pete was created and what makes it different from a general-purpose AI tool. Pete combines the capabilities of a large language model with Catholic guardrails and more than 20 years of Petrus fundraising experience.  We discuss how Pete uses the knowledge and experience built into it, how it is designed to remain grounded in Catholic teaching, and how it can learn about an individual organization’s mission, history, and fundraising needs.

 

We also address what Pete is not. It is not a fundraiser, a decision-maker, or a replacement for human relationships.

 

Its purpose is to help Catholic fundraisers work more efficiently so they can spend more time with donors, leaders, and the people their ministries serve.Understanding the tool’s strengths and limitations is essential before relying on it.

 

Episode Transcript:

Matt: I love seeing how ministries that maximize their fundraising are changing the world. Being able to build Pete to help our fellow fundraisers and our peers do that is incredible.

Matt: Welcome back to the Catholic Tech Table podcast. My name is Matt Bond, senior consultant for Petrus Development, and I'm joined today by none other than Miss Jenni Reimer. Jenni writes and teaches fundraising courses for us at Petrus, as well as many other wonderful things.

Jenni: Thank you, Matt. I'm so glad to be on this show again. We've been having a lot of fun, and we're going to have a good discussion today.

Matt: Very good. If you're unfamiliar with Petrus Development, real quick, we're a Catholic fundraising company. We help teach fundraisers how to fish, not give them the fish. Since 2004, Petrus has helped raise millions of dollars for Catholic nonprofits that are building the kingdom of God. In this podcast today, we're discussing how we can use AI to help us build every element of an annual giving program.

Matt: We're specifically using Pete, which is Petrus' very own AI, created for Catholic fundraising experts.

Jenni: That's right. The way we're going to show you how you can build the annual giving program really quickly by using Pete is by creating this concept of a guinea pig organization. We invented a fictitious organization, which we call the St. Jude Catholic Leadership Center, and this is going to be our demonstration organization.

Jenni: On the Tuesday episodes, we do a practical how-to of some aspect of creating, maintaining, or thriving in a fundraising program. Then on Fridays, we're going to step out of that practical, tactical view and talk about the theology and the ethics of using AI as Catholic fundraisers. So today's core question for our discussion is: What is Pete, and why should we trust it to help us do our work as Catholics?

Jenni: But before we get into that, let me share really quickly that this episode is sponsored by The Harbor. The Harbor is Petrus' Catholic membership community. It's an online subscription, and Petrus created it for people who are trying to raise money for Catholic organizations. It gives fundraisers the tools, guidance, accountability, and support that they need to succeed in their jobs.

Jenni: Membership includes the professional level of Pete AI, so you'll have access to the same tool that we are using throughout the show. Let's get started. Matt, you were instrumental in building Pete. Would you mind sharing with us a little bit about how it got started, how you were inspired to create it, and the background story?

Matt: Yeah, it's kind of an interesting background story. AI has been around since the '80s with Terminator and the stories behind it. But just to put that fear behind you, we're a long way away from that, and I don't even know if that's a realistic expectation of where AI could ever possibly go.

Jenni: I hope not.

Matt: But I say that jokingly. And, Jenni, I see you rolling your eyes over there.

Jenni: No, I only do because I remember the moment years ago when my husband said, "Hey, AI was launched and it's having conversations with people." And I was like, boop, boop, boop, Terminator immediately popped up. So I can relate to that analogy.

Matt: I just recently watched Jarvis. No, that's not it. Well, Jarvis, of course, is Iron Man's assistant, very much an AI. We're not there even yet, at least not in the commercial use version of it. But we do see that with Ultron being the bad guy, and Vision being the good guy of the AI versions of that too.

Matt: And so we're probably closer to that level.

Jenni: I'm so glad. I like that comparison much better.

Matt: Yeah, except, again, both of them are thinking on their own, and we're a long way away from that. There's one thing to know everything and then be able to put pieces together, but the other thing is to take this thought and this thought, put them together, and start reasoning out multiple thoughts of your own, as human consciousness goes.

Matt: So enough about that. That's pipe dreams, and we're dreaming a little bit about the future. But the reality is AI started turning the corner, and we realized that it was going really fast. With that, we thought, "Oh, hey, this will be fun," you know.

Matt: Andrew always tells a story. He started doing drawings with AI for his kids. For me, it wasn't even that. I actually started at a professional level. I was struggling writing a course, the major gift course, putting all my thoughts together because they were so scattered around.

Matt: I remember Andrew and Rhen just saying, "Hey, have you put your thoughts into AI to help you put them in an order that makes sense?" So I did that. Totally mind-blowing, what came out. Better than anything I could have put in, not creating my thoughts, but organizing my thoughts into one coherent thing.

Matt: That led to us getting together in September of 2025 to do what's called a hackathon. There were four of us: Andrew, Rhen, our old intern named Josh, who's moved on to wonderful things, and then myself. We got together for two days and just said, "Let's build whatever we can build."

Matt: We went through there and built amazing, crazy things, and hit our heads against the wall a few times. Fast-forward to about the March timeframe: what took us two days in March, through the AI advancements, we were able to do in about 30 minutes, come March. When I realized that we had jumped that far, I had the crazy idea of, "Let me just see if we can build Pete.

Matt: I wonder if it's possible now, where we're at." So I jumped in and started getting to work with Pete, building kind of Pete 1.0. We have a lot of dreams for Pete, but what you see today is Pete 1.0. We turned around and built 1.0 in the course of about a month and a half of some serious effort, and then testing and whatnot.

Matt: We're still adding and building upon that. But that's really the... not the why, but the what we built and how it came to be. The why behind it is we saw the fruit that could come from it, but we got scared of where it came from. We knew that there needed to be a place that could be trusted, that people knew would give Catholic answers and would be theologically Catholic at its core, so that you could go to it and have faith that whatever came out of it would not be contrary to our belief and our faith.

Matt: That was a big piece of it. We thought, if we don't get out in front of this, it's either going to get out of hand or somebody else is going to build it. Then we thought, "Well, if somebody else is going to build it, why don't we go ahead and build it?" So we jumped off that ledge to help our fellow fundraisers out there do their mission and their good work.

Matt: Quickly, what Pete is: you've heard of LLMs, you've heard of ChatGPT, Claude, Gemini, Grok, Perplexity. All of those are LLMs, which stands for large language models. What an LLM really is, is what we think of as AI intelligence. Imagine going through your school library or your city library and reading every single book there, but also being able to recall everything that you've read. That's really what a large language model is, but on the nth scale.

Matt: It's not just your library; it's as much as they could possibly put into it, they did. Then it remembers that, and it's able to pull those things out. When you know so much, you're able to finish sentences, think logically, and pull up how this person did it and how that person did it, and put not two different thoughts together, but the same thought together in what you're using.

Matt: Kind of like following a recipe, for example. And we'll talk about RAGs and other stuff, which is also your main form of memory. But if you know all of this, you become a little smarter and are able to finish sentences. If I said "peanut butter and," we all know it's "jelly," and it's the same way with an LLM.

Matt: They know that the next answer is "jelly" in the same form. That's a very simplified version of explaining what an LLM is. It's just a really smart friend that remembers everything it read and can recall everything it's read, but can also put those pieces together and predict what the next word and sentence is going to be.

Matt: How that turns into logic is above my pay grade, but that's definitely where it comes from. Then, if you look at where Pete goes into it, Pete has all that intelligence, but we put bumpers and guardrails on it so that it follows where we want it to go. Meaning we say, "This is Catholic in its entirety, and this is what you use.

Matt: This is how fundraisers work." So it gets all of that, but... Think of it as... we were joking earlier that I had to chase my dogs around this morning because they got out. So we'll use a dog example. ChatGPT and Claude can go out and train a dog, but you've got to learn the dog's name.

Matt: You've got to learn its quirks and whether it's food motivated and all those different things that you learn with training a dog to sit or to roll over or whatever it is. What we do is we come in and give it all that information, so it knows... I say "give it to," but it knows the dog's name, it knows how it thinks, it knows that it's food motivated, it knows that it can roll over and it can do this.

Matt: From that, it can use that data to help train that dog to take the next steps in the direction that we want it to go. That's the difference between Pete and Chat. With that, next week we'll go into RAG, and that's going to explain in a lot more detail exactly how those guardrails work and how it focuses specifically on Catholic fundraising.

Matt: But that's just a quick synopsis of how an LLM works.

Jenni: That's awesome. I appreciate that. You broke it down in a way that I can understand. I'm not a technological-background type of person; I just appreciate being able to use the technology. So, all right, we covered what an LLM is and how Pete is different from something like ChatGPT, but how is it different from a search engine?

Jenni: Is it just a glorified search engine?

Matt: Mm, it can be used as a glorified search engine, but it's not a...

Jenni: Just the surface, that's...

Matt: Yeah. It's not necessarily like a search engine. With a search engine, you put in what you're looking for, and it pops up everything that has anything to do with it. As we've used it, if you type in Nike basketball shoes, it's going to pop up Nike basketball shoes, but it's also going to pop up Michael Jordan wearing Nike basketball shoes, and it's going to pop up an eBay browser that's selling Nike.

Matt: So it's going to pop up all these different pieces, and you have to be the one who searches through that to find the relevant information. With an LLM, when you ask it that specific question, for example, it's going to do something similar, not through a search engine, but something similar.

Matt: But it goes through and filters out all the stuff that has nothing to do with what you're talking about, to help you specifically answer the question that you asked.

Jenni: So the output is much more tailored to what I was looking for and not just a general catch-all.

Matt: Correct. Yep.

Jenni: Awesome. That sounds good so far. So how this is going to play out in our main project, which is going to be building a development program: we're going to use Pete to help us move faster and be more effective at all of the different...

Jenni: At creating all the different aspects of the development program, specifically the annual fund to get started with. We're going to show the prompts that we use and then the output that Pete gives us. These prompts that we're working with are going to get better, and we're going to figure out how to use it better and how to get an even more effective output from it.

Jenni: We're also... On these Friday episodes, we're going to question, well, when do we go too far? How is it a good use? How is it not a great use? The point of the show isn't to make it seem like Pete can do everything, because we're going to run into limitations.

Jenni: But the point is to see how we can genuinely make our work as fundraisers more effective and do it faster so that we can get back to the relationship-building aspects of our job, which is ultimately where the kingdom of heaven is built from that aspect of our job. So I think that was a great background.

Jenni: In the next Friday discussion episode, as Matt mentioned, we're going to do a deeper dive into RAG, which is where we'll get into some exciting and specific elements of Pete. But I think that was a great intro, Matt, and I really appreciate you joining us and giving us all that background and that technological expertise.

Matt: Thank you. I appreciate that. It's a passion project for me. I love fundraising. I love seeing how ministries that maximize their fundraising are changing the world. Being able to build PEAT to help our fellow fundraisers and our peers do that is incredible.

Matt: It's a wonderful opportunity, and I'm super excited about it. I'm looking forward to next week.

Jenni: Yeah, that's right. And so... I got ahead of myself too. That's our Friday episode, but Tuesday, we're going to get started on how you build your case for support with AI and how Pete is going to help us with that. I know that "case for support" can be a phrase that causes some anxiety for people, so we're going to jump right in on that, and I will see you all next episode.

Jenni: Thanks so much for joining us.

Matt: Looking forward to it.