How We Think About AI

AI is fundamentally a pattern-recognition engine. It’s good at reading large volumes of text and identifying structure: recurring themes, gaps in coverage, how often certain topics or passages appear, whether a sermon’s introduction and conclusion connect. It can compare a draft against a body of past teaching and surface observations a human might miss — not because it’s smarter, but because it can hold more information in its head at once.

These are the same kinds of things a well-read preaching coach or a diligent seminary professor might notice if they had the time to review every sermon you’ve ever preached. AI just does it faster and more consistently.

Sermon Analyst uses AI in this way: to aggregate, summarize, and find patterns in your preaching, and sometimes to provide additional context based on its training data. Every report, every analysis, every piece of feedback the system produces is grounded in what your church has actually taught — your sermons, your stated goals, your context.

AI cannot discern spiritual truth. It cannot be convicted by a passage of Scripture. It cannot sense that a congregation needs to hear something difficult, or that a sermon needs to linger on a moment of grace rather than move to the next point. It cannot pray.

These aren’t limitations that may be solved with better models. They’re category distinctions. The work of the Holy Spirit in sermon preparation is not an information-processing problem, and treating it as one misunderstands both the technology and the theology.

This is why we don’t build features that generate sermon content, suggest what you should preach, or offer theological opinions. We think this would be a mistake — not because the output is always bad, but because the process matters as much as the end result.

Pastors have long used tools in sermon preparation: concordances, commentaries, search engines, the writing of other human beings. This is how we see Sermon Analyst — as a reference tool, not a ghostwriter.

The distinction we maintain is this: the pastor is the author. The pastor chooses the text, develops the theme, structures the argument, and decides what to say. Sermon Analyst provides data and observations that the pastor can use or reject. When our analysis shifts from reporting patterns to offering interpretive commentary, it says so. The system is designed to inform your reflection, not replace it.

We think this is the right line for a tool serving pastors. But we also recognize that thoughtful people will draw it differently, and we respect that. If you’re a pastor who’s uncomfortable with any AI involvement in your preaching process, we understand. This tool may not be for you, and that’s fine.

One thing that distinguishes Sermon Analyst from a general-purpose chatbot is that every input is explicit. The system analyzes your sermon library, your church’s stated context, and whatever you specifically type as a prompt. That’s it.

It doesn’t learn from your conversations. It doesn’t adapt to your feedback over time. It doesn’t build a hidden memory of your preferences and subtly shift its analysis to match. Every time you run a report, the system approaches your content fresh, using the same consistent framework.

This is a deliberate design choice. A tool that gradually adapts to your biases can end up reinforcing them — telling you what you want to hear rather than what might be useful to consider. By keeping the inputs explicit and the process transparent, we make it easier for you to evaluate what the system is actually doing and decide for yourself how much weight to give it.