Most of what limits a church’s digital ministry is not money and it is not talent. It is that somebody has a calendar with a fixed number of hours in it, and that person is already spending all of them.
That is the thing worth saying first about AI in ministry, because it decides which uses matter and which are noise. AI is genuinely useful to your church in the places where a human being had run out of hours. Everywhere else it mostly helps you produce more things that nobody was waiting for.
This is the first in a weekly series about that — how AI multiplies the impact of a church, and specifically what it lets one person or one small church now do that they could not do before. Not whether AI is coming. It came.
Where the ceilings actually are
Go through your digital ministry and ask which parts stop growing when one person gets busy. The list is usually short and always the same.
Questions. Somebody asks your church something real at 11pm. Whether they get an answer depends on whether a specific tired person had capacity that night. Multiply by every channel you are on.
Follow-up. Someone watched an entire service online for the first time. In almost every church I know, nothing happens next — not from indifference, but because the person who would do it is already doing four other things. The people who came once and were never contacted are the largest group in most churches’ digital ministry, and nobody has a plan for them because there has never been time for one.
Language. Your church speaks one language, usually. The people who found you might not. That gap has historically been closed by finding a bilingual volunteer, which means it is closed only where such a volunteer happens to exist.
What you already said. You have years of messages, and functionally none of it is findable. When someone asks a question you answered thoroughly in 2021, nobody can locate it, so the answer is either retyped from memory or not given.
Every one of those is a capacity problem wearing the costume of a strategy problem. And every one of them is where AI does something real — not “writes your sermon,” which nobody asked for and which is the wrong use anyway.
The multiplication test
When you are weighing whether an AI tool is worth the trouble, one question does most of the work:
Does this let a person do something they genuinely could not do before — or does it only help them make more of what they were already making?
The first is multiplication. The second is output, and output is not a ministry goal. It is very easy to use AI to triple how much your church publishes and change nothing about how many people are actually reached, which is roughly what happened to a lot of organisations in the last two years.
Concretely, on the same side of the line: transcribing every message you have ever preached so the answers in them become findable. Translating a follow-up message into the language someone actually reads. Drafting a first reply to forty questions so a human can correct and send them in the time it used to take to write four. Turning one message into the handful of formats your people genuinely use.
And on the other side: generating social posts because the calendar has holes in it. Writing devotionals nobody requested. Producing a newsletter that exists because newsletters exist. All of that is easier than it used to be, and none of it lifts a ceiling.
The empowerment edge makes the same argument without the technology: a ministry that depends on one person’s availability has a ceiling regardless of how gifted that person is. AI is one of several ways to raise it, and it happens to be the one that arrived this decade.
The part that stays human
There is a line and it is not subtle, so I will state it plainly: AI can help you prepare a relationship. It cannot be one.
Drafting a reply is fine. Translating it is fine. Finding the thing you said three years ago is fine. Sending an AI-written message that implies you personally thought of somebody, when you did not, is not — and people work it out faster than church leaders expect. The cost when they do is not embarrassment, it is that they stop believing any message from you.
So the rule I use is that a person stays behind every relationship. AI does the part that was never the relationship anyway: the transcribing, the finding, the first draft, the translation, the sorting. Our own AI guide works through the harder cases — the uses that need actual theology rather than a policy — and those are worth reading before you get far into this.
What I would do first
One thing, this week, that costs nearly nothing:
Transcribe your last three months of messages and read what you have.
Not to publish it. To find out what is in there. Almost every church discovers that it has already answered the questions people keep asking — thoroughly, pastorally, in the pastor’s own voice — and that the answers have been sitting in video files where nobody can search them. That is not new content. It is content you paid for years ago and never took delivery of.
Then ask the question this whole series is about: now that this material is findable, what becomes possible that was not possible last week? For most churches the honest answer involves somebody finally being able to answer a stranger at 11pm with something true that their own pastor already said.
That is multiplication. It is not very exciting, and it is worth more than anything you will generate.
Part of the weekly AI series. The full guide is AI for churches.
