Before we go further, let’s clear the air.

There are three things I hear almost every time I talk to church leaders about AI — and all three are wrong.

The first is that AI is just a glorified search engine or chatbot. It isn’t. A search engine retrieves information. AI possesses the sum total of human knowledge and has the ability to connect any piece of that information with any other — nearly instantly. It can take that knowledge and create something new and novel. It can pivot an old theme into a new concept. It can execute tasks. Based on the result of an evaluation it can make autonomous decisions and work independently with near zero oversight. It doesn’t sleep, eat, or get tired. It can execute a near endless number of tasks in parallel.

That is not a search engine.

The second misconception is that deploying AI — especially an agentic solution — is too hard or too expensive for a church. We’ll come back to this. But the short answer is: it was. It isn’t anymore.

The third is the one that stops more conversations than any other. AI means replacing staff. AI means firing people. I want to address this directly, because it’s both understandable and wrong. The goal of AI in a ministry context isn’t to replace your people — it’s to free them. There’s a difference, and it matters enormously. We’ll come back to that too.

The Problem We’ve Always Had

In the church world — especially the megachurch world — we’ve always desired to connect people to other people. The problem is that process doesn’t scale. At all.

Different people have different needs, different circumstances, different constraints. Not all people fit with all other people. Whether you’re trying to build a group, connect someone to benevolence resources, or find the right serve opportunity — you need a process to sit between the people. And so we built one. And then another. And then another on top of that.

Process on process on process has a name. It’s called bureaucracy. And bureaucracy is slow, impersonal, and almost always misses the original intention of what you actually wanted to do in the first place — which was just to connect people with people.

Think about the early church. Even they had processes. Acts 6 gives us a glimpse of the early food distribution system — a genuine logistical operation designed to make sure widows weren’t overlooked. The process existed to serve the people, not the other way around. Somewhere along the way, for many of us, that got flipped.

The flow became: People → Process → People.

And in that middle layer — that well-intentioned, carefully constructed, increasingly complex middle layer — we lost people.

What AI Actually Changes

Here’s the shift that AI makes possible: we can now automate the process layer almost entirely — and in doing so, get back to what we actually wanted, which was people connecting with people.

Let me give you a small but vivid example.

I serve on the next steps team at my church. One of my jobs is to meet people and connect them to the baptism process if that’s their next step. Simple enough. Except — I often forget where the form is. Or exactly what information the form needs. Or who they actually need to talk to next. (Spoiler: it’s supposed to be me. And sometimes I forget that.)

What ends up happening is I connect this person to the process. They fill out a form. They answer a series of questions. That goes into an email. Someone from the staff has to get back to them. A conversation gets scheduled. We make sure they’re really ready for this next step. We schedule the baptism. We get the right shirt size. And so on.

Now imagine a world where AI never forgets the process. Always asks the correct question. Knows everyone’s calendar. Immediately connects this person to the right individual — and provides that individual with exactly the information they need to have the right conversation.

That’s not a fantasy. That’s an agent.

What Is an Agent, Exactly?

When I share with a large group I use this illustration.

If generative AI is like a calculator — useful for solving particular problems, waits for your input, provides an answer, presuming you pressed the right buttons and asked the right question — then an AI agent works like a general contractor.

A general contractor knows the schedule. Knows all the moving parts. Knows how to coordinate when supplies are going to be dropped off and who needs to do what next. They navigate when things don’t go as planned. They surface decisions that need to be made — but mostly, they manage the whole process of building your house.

AI agents work similarly. Given access to the right tools and information, they work autonomously toward your objective. They don’t wait to be asked. They execute.

Grace

I had the opportunity recently to speak with the Lead Pastor of a large church in Toronto. He told me about Grace.

Grace is the receptionist for his church. She’s kind. She’s authentic. She’s well-informed and even empathetic. She knows exactly who to connect you to when you call in. You don’t get the runaround of menu options — she simply listens to what you have to say, understands your situation, and connects you to the right person.

Grace is an AI agent.

She’s not really a “she” at all — she’s a purpose-built implementation of generative AI designed to understand the church’s current programming, what’s available, who’s available at any given time, how to reach them, and how to map people directly to people. She doesn’t pretend to be human. But she makes interacting — especially in a stressful situation — easier. Her entire goal is to make anyone who calls feel comfortable, cared for, and ultimately connected to a real person.

This church (who I hope to do a profile piece with) is on the leading edge of using AI in a ministry context. But I share Grace’s story because she represents something that’s possible right now — not in five years, not when the technology matures. Now.

And she never takes a lunch break.

The Tightrope

This is where I need to be honest with you, because I’d be doing you a disservice if I only told you the upside.

This is a tightrope. On one side — not enough automation — and you end up doing work you didn’t need to do, missing people in the process. On the other side — too much — and you compromise the very thing you were trying to create. Connection.

Here’s the guardrail I’d offer: when your AI efforts stray into creating unintentional disconnection — or worse, artificial connection — that’s where the real risk lives.

Take Grace. If Grace offered to pray with someone who called in distress, that would immediately feel inauthentic. From a technical standpoint it’s entirely possible. Chances are AI could be configured to pray the world’s most theologically precise prayer. But just because you could doesn’t mean you should. That’s where discernment comes in. You know your people. You know your culture. You know what will feel off. And it’s going to be different for every church.

AI — like all things — is another moment for us to be as wise as serpents and as gentle as doves.

Give yourself grace in this process. Pun intended — because you could actually implement Grace yourself. Not everything you try is going to be a win. Sometimes you’ll push the boundary too far. Sometimes you won’t go far enough. That’s not failure. That’s learning. And in this moment, churches that are willing to learn are the ones that won’t get left behind.

So Where Do You Start?

Same answer as last time — start simple, but simply start.

Pick one process in your church that exists purely to move information from one person to another. A connection card follow-up. A volunteer scheduling workflow. A first-time visitor response sequence. Something where the human work is mostly administrative and the human value is mostly relational.

Now ask: what if AI handled the administrative layer entirely — and your people only showed up for the relational moment?

That’s the shift. That’s the mandate. And the tools to do it exist today, at a price point that would surprise you.

People to people. That was always the goal. AI just finally gives us a way to get there.