AI Fundraising Build Pt. 6: What Is RAG, and Why Does It Matter?

 

Episode 11, October 9, 2026

 

 

Show Notes:

Why does Pete give different answers from a general-purpose AI tool, and how does it know what information to use?

 

In this episode, we explain RAG, or retrieval-augmented generation, in practical language.  We discuss why this matters for accuracy, consistency, organizational voice, and Catholic guardrails.  You’ll also learn how organizations can use their own case for support, appeals, reports, and other documents as reference material.  Most importantly, we explain why RAG does not make AI a replacement for a fundraiser.

 

The goal is simple: let AI handle more of the work that keeps fundraisers at their desks so you can spend more time doing the human work of fundraising.

 

Episode Transcript:

Matt: Welcome back to the Catholic Tech Table. My name is Matt Bond, and I’m here with the wonderful Jenni.

Jenni: Hey, Matt, how are you doing?

Matt: I’m wonderful. How about yourself?

Jenni: I’m great. I’m glad to be on the show today. Excited to be talking more about how we can use AI in our Catholic fundraising work.

Matt: That makes two of us. We’re here with Petrus Development. If you’re unfamiliar with Petrus Development, we’re a Catholic fundraising consulting firm. We take a teach-to-fish approach, coaching clients to better their ministry through financial funding. We help Catholic organizations build the skills, strategies, and systems to develop sustainable fundraising programs.

Through that, one of our great projects is the Catholic Tech Table, where we are building a fundraising program for St. Jude, which is completely fictional. We’re doing it one piece at a time, using AI along the way. We’re actually using Pete AI, a tool built specifically for Catholic fundraisers. Today, we’re going to explain one of the things that makes it different: RAG. That’s R-A-G, RAG.

Jenni: Up to this point, in our last demonstration episode, we used Pete to help build St. Jude’s case for support. We took stories and content about how St. Jude has been changing lives and doing good work, and we gave it to Pete. Pete organized it into a compelling case for support, or at least got that process started for us. That took about 20 minutes, which is a pretty incredible win for St. Jude and for Pete.

You can get Pete by purchasing a monthly subscription, or you can join the Harbor, which is our sponsor for today’s episode. The Harbor is a monthly membership community from Petrus Development, designed to help people working for Catholic organizations work together in community and raise money more effectively. It’s designed to give Catholic fundraisers the support, guidance, accountability, online tools, and more that they need to keep moving forward in their jobs. And as I mentioned, membership includes the professional version of Pete AI, so you can access the same tool that we’re using throughout this podcast and follow along with us step by step.

Today, as Matt mentioned, our core question is: What is RAG? How does it change what Pete is, and what difference does it make when we’re asking Pete to help us create fundraising content? Now, full disclosure, I know very little about how AI works. When I was just using AI for my personal life, planning meals for the kids and researching how to grow better vegetables, I was fine with that. But now that we are using AI in the work of the Holy Spirit, I need to know more about how it works. I need to understand some of the background. I certainly don’t need to start following all the different tech podcasts that are out there, but understanding what it is as a tool is very helpful to being able to use it effectively, carefully, and responsibly in our work.

Matt, you’ve been part of the Pete AI creation process, as chief architect of the entire project. In our last Friday episode, we discussed why you made Pete, and we skimmed the surface of what it is and how it works. Can you start this conversation by sharing how Pete is different from a regular large language model like ChatGPT or Claude?

Matt: Yeah, sure. The big difference is going to be RAG, which stands for Retrieval Augmented Generation. It’s how Pete retrieves information, but also how it stores information, where it stores it, and what it stores. The biggest thing we have in place here is how we’ve trained it.

We’ve taken the past twenty years’ worth of documentation, fundraising talks, classes, courses, and support, everything Petrus has done over the past twenty years, and we’ve stored that within Pete to use, along with other pieces we’ll get into in just a moment. What gives Pete that Catholic sensibility is that we put up guardrails so that it can only use that information. That’s all it can pull from to produce meaningful content. Someone might ask, “Why don’t you go for more?” But we want to make sure it’s one hundred percent authentically Catholic.

As we started Pete, the big question was whether we could trust the content that comes out. It has to be within the Church’s doctrine and the Church’s teaching. It has to be good, solid content. So we placed guardrails from the very beginning. Before we even started thinking about building Pete, we created an ethics sheet: What ethics will Pete run on, before anything else happens? That was the beginning, because it was going to be a very fast-paced race to the top, as we’ve all seen with how quickly AI has grown.

So that’s a quick difference between Pete and what you’re going to see in Claude, Gemini, Grok, and ChatGPT, the big four.

Jenni: It’s really a dynamic. We are Catholic fundraisers, and we want to use AI in a responsible and ethical way. So now I have a tool in Pete that helps me do that, and we are working together. The robot and I are on the same team to make sure that what we’re doing is grounded not only in good fundraising practices, but in Catholic teaching.

Matt: Exactly. And as we go along, we continue to add more information. Let’s jump straight into RAG, Jenni, instead of going around it. Think about grocery shopping. Where I’m from, Walmart’s the big grocery store. I’m a big fan of home delivery, although it does cost, so we usually go pick up. Either way, it’s great. But think about bags of groceries being delivered. Inside the bags are each ingredient: your apples, your flour, your sugar, your milk. All of that comes in the bags after you go grocery shopping. And for the rest of the week, that’s all you have to use to make a meal.

Translate that to what we’re doing today with Pete AI. We’re giving it specific bags of groceries, and we’re using the intelligence of the large language model, Claude specifically. It may change in the future, but today it’s Claude. We’re using its thinking ability, how to write a sentence, but only giving it the ingredients to write from, the bags of groceries. That’s part of what RAG is: the memory side of it.

One of the beautiful things within Pete is that there’s more than one type of bag of groceries you can use. You have the Petrus bags of groceries I referred to. But as we all do, we don’t just shop at one store. Sometimes we go to Whole Foods, and sometimes we go to Aldi. We like their fruit from Aldi, so that’s another bag of groceries. Within Pete, you’re able to upload your own documents, such as your case for support, your end-of-year letters, and your appeals. You’re able to upload those into Pete, and only you can use them. Only you can use them. I want to repeat that. Only you can see and use these documents to create your own voice and to start creating your own bag of groceries for your specific organization.

Jenni: So that information is not being sent to whatever language model we’re using, whether it’s Claude or ChatGPT. The information from your own organization is not being sent to teach the AI.

Matt: Correct. It’s not there to teach the AI. It’s just used as another ingredient.

Jenni: Your pantry.

Matt: In your pantry. We’ll talk about retrieval in just a second. The third piece is projects, which we’ll touch on briefly. When you create a project, you get to create a small bag of groceries that you can build from as well. So there are multiple bags.

Lastly, we’re continuing to build the Catholic knowledge base that Pete has access to by pulling in encyclicals, Church teachings, and things of that nature. That way, you can tie not only the fundraising side but also the theology side to some of your answers and some of the things Pete is able to produce. So that’s your entire pantry, to use the word you used, and that’s all Pete can use. He can’t pull any other ingredients from anywhere else.

On top of that, and speaking of ethical things, here’s another fun thing. My family is not big on red sauce. Myself and a couple of my boys, although we like red sauce, red sauce doesn’t like us, so we literally can’t and shouldn’t eat it. Likewise, we have rules in place so that even though you have all those groceries, and they are Catholic, you can’t use those ingredients to create something that doesn’t bode well for the family, so to speak. Meaning, it can’t go against the Catholic Church and its teachings. So you can put all that information in there, but if you ask it to write an end-of-year letter for a pro-choice organization that you volunteer for, it’s not going to write it for you.

Jenni: Hmm. That’s really interesting. So it knows Catholic teaching, it knows your organization, and it knows the last 20 years of Petrus fundraising success.

Matt: Mm-hmm.

Jenni: I mean, this is sounding pretty good. Is this going to replace Catholic fundraisers in the future?

Matt: What’s funny is that, as we record this, a few days ago the Cardinals got together in California and met with a lot of the architects of this technology to talk, and they were given dire news from these individuals. I don’t fall into that realm. Here’s the thing: we are human. At our very nature, we are human. And technology, no matter how hard we try, cannot replace human interaction. An AI robot cannot hold a baby and have the baby think it’s his mom. An AI robot cannot console a grieving spouse. It cannot discipline a child who’s acting out. It can try, but it can’t get the human side of things. In the same way, whether over the phone or otherwise, an AI cannot replace sitting in front of someone, sharing your mission, and asking them to support it. So can it make your life easier, along with everything else it does? Absolutely. But it cannot replace that.

Jenni: Great. So it’s not my competitor, it’s my assistant.

Matt: That’s right. That’s a great way to look at it.

Jenni: In one of our previous discussion episodes, Andrew referenced a beautiful quote from the Catechism of the Catholic Church, which I have written here: “The human person participates in the light and power of the Divine Spirit.” To me, that sums it up. No computer system, even one as well trained and educated as it might be, is going to be able to participate in the light and the power of the Divine Spirit. We are a tabernacle of the Holy Spirit.

So how can these wonderful, sophisticated education systems enable our work to be more effective? How can Pete unlock my full potential as a Catholic fundraiser to build up the Kingdom of God and minister to His people? I think it’s just the beginning to see where we can go. We’re working through this in this podcast step by step, and we’re really figuring it out together. We’re just skimming the surface, and I’m excited to see the potential of this really wonderful program that Pete is, and to see how it can make a difference in our fundraising lives.

Matt: Absolutely. It reminds me of the reason I got into helping build Pete. I went into a number of ministries that didn’t have enough funds to support the staff they needed to get their good work accomplished, so things had to be cut. They weren’t able to spend as much time going on face-to-face visits, sitting down with donors, finding out what makes them tick, what they love, and why they support the organization, and continuing those relationships. Instead, they were spending a week writing an end-of-year letter because it was so important to get it right, or two days writing their annual funding plan. Those are just two items on a litany of things a development director has to produce every year to run a successful program. But it took away the thing they needed to do the most.

So it was twofold. You had entities that knew they needed to raise more funds but couldn’t, because they didn’t have the funds to hire the staff to do the work. And then you had the staff, but they had so much on their plates that they couldn’t do the things they needed to do to be successful.

My driving force when Pete came out was: How can we solve this so that a development director, an executive director, or a priest can sit down with donors and do the things that help the ministry move forward, instead of the things that have to happen behind the scenes for the other pieces to work? How do we reduce that time, or alleviate it, so that they can make the most of the time they have to support their mission?

Jenni: Yeah. Beautifully said, and that tees up our next episode very nicely, because we’re getting into data entry: how do we get messy spreadsheets into a central database, a CRM that’s going to enable our entire development program to work and build fruitful, authentic relationships? Data entry is always at the bottom of everyone’s list of things they want to do, but it’s so essential and important to our long-term relationship-building process. I’m looking forward to seeing how we can use Pete to help with that challenging process, which, like you said, keeps us away from people because we are searching through boxes of old files, trying to find all the people who have been lost over the decades of good work our nonprofits have been doing. Thank you so much, Matt, for sharing all that insight and your experience building Pete, and for helping us understand a little more about why it’s an effective tool.

Matt: Absolutely. My pleasure. I hope I didn’t nerd out too much and put too many people to sleep, but it’s something I’m very passionate about, so thanks for letting me share.

Jenni: Absolutely. Thank you so much. If you’re not subscribed, subscribe to the show, and join us on Tuesday, when we’ll get into some data processing.

Matt: Great. Can’t wait.

Jenni: Thanks, everybody. Have a great day.