AI can now do a lot of the legwork around a sermon. It can gather background on a passage, raise questions you did not think to ask, find the paragraph where your outline loses people, and argue with you like a skeptic in the back row. It cannot do the one thing a sermon is for, which is to bring a word to these particular people. That part stays with you.
The pastor with no staff is the one who gains the most from the legwork.
Who helps a small-church pastor prepare?
Usually nobody.
A pastor in a church of sixty has no research assistant and no editor. No one to say “I got lost in the middle” before Sunday instead of after it. You read what you can, you write at the kitchen table, and you stand up and find out in real time whether it worked.
Plenty of faithful pastors preach that way, week after week, and never once get a second pair of eyes on a message before they give it.
That is the gap AI can fill, if you let it do the right jobs.
What can AI actually do for sermon prep?
Three things, and they are all prep.
It can be a researcher. Ask it for the historical setting of a passage, what kind of letter or story you are looking at, who the first listeners were. Treat the answers as leads to check, not as facts to repeat. A machine will state a shaky claim in the same confident tone as a solid one, so anything you plan to say out loud, you verify.
It can be an editor. Paste in your own draft and ask where it is unclear, where it repeats itself, where a listener might drift. You will be annoyed by about half of what comes back. The other half is the note a good editor would have given you if you had one.
And it can be the skeptic. Tell it to respond as a thoughtful person who is not sure they believe any of this, and ask what that person would push back on. The objection you would never have raised yourself is often the one sitting in your third row.
This series keeps asking one question: what can one person, or one small church, do now that they could not do before? Here the answer is that one pastor can prepare with the depth that used to need a team. The AI for churches guide shows where else the same thinking applies.
Should AI write the sermon?
No.
I’ll confess the temptation is mine too. A polished message assembled from other people’s material is the easy road, and I have wanted the easy road more than once. It would be clear, organized and even funny. And it would cost me nothing.
I should name that plainly. A sermon that costs the preacher nothing tends to land on the listener the same way.
So let me ask the questions I ask myself. Did I pray over these actual people? Do I know who is sitting in that room who just lost a job? Did I wrestle with this text until it wrestled back? Could I look the congregation in the eye and say “this is what I believe,” and mean every word?
A generated sermon answers no to all of them. It has never met your congregation, it has never been wrong about God and been corrected, and it will not be sitting with the family after the service.
AI can prepare a relationship. It cannot be one.
Can I let AI quote Scripture?
Do not.
This is the rule I would put on the wall. Never ask a model to produce a Bible verse for you and then trust what comes out. It can mix translations, blend two passages together or hand you wording that sounds right and exists nowhere. The more familiar the verse, the more convincing the error.
Pick the translation you preach from, copy the verse straight from it, and check the reference yourself before it reaches a slide or a manuscript. That takes a minute. A misquoted verse said from the pulpit can take a lot longer to live down, and it teaches your people that the text is approximate.
Scripture is not something to generate. It is something to receive.
What about the rest of the week?
The hours you save should go somewhere on purpose.
If the research and the first edit take less time, you now have room to do the thing a tool cannot. Call someone. Sit with the person the text made you think of. Read the passage out loud on a walk. Thirty minutes of preaching used to be built on far more hours than most people in the pews would guess, and a lot of those hours were hunting for background, not thinking about the people you were preparing it for.
This is where the empowerment edge comes in. If you have someone in your church who loves to dig into a text and has never been invited to, this is a way to let them. Give them a passage and a few good questions. Let them bring you the background and the objections, and then you do the praying and the shaping. A tool did not replace your people there. It gave you a reason to ask them.
One thing to try this week
Take a sermon you have already preached. Paste it in and ask for the three places where a listener who is skeptical would push back, and the one place where someone might have gotten lost.
Read what comes back. Don’t change the sermon yet. Just notice what a fresh set of eyes sees that you could not, and decide which parts you would want to hear before next Sunday and which are noise.
So let me ask you, Pastor. If you had one honest reader in the room before you preached, what would you want them to tell you?
Try the tool on a sermon you have already given this week. Then call a person in your church who would be willing to be that reader every Sunday.
Part of the weekly AI series. The full guide is AI for churches.
