Automated sermon study notes
Sunday's sermon, still working on Thursday.
Every week, a church's new sermon is transcribed and edited into a five-minute study note — what was claimed, why the text supports it, and what to do about it. It is waiting before the week starts. Nobody types a word.
The gap
The recording of last Sunday's sermon is one tap away, and almost nobody plays it. Not because the message didn't matter — because re-listening costs forty minutes to recover something you already half-remember.
The gap is form, not content.
Why it isn't a summary
A summary compresses. This is edited.
Organized by argument
Headings name the idea the preacher was making, never the story that delivered it. The note follows the shape of the reasoning, not the order of the talk.
The scripture is verbatim
Verses are printed from a vendored Easy-to-Read Version dataset, not reconstructed from a transcript. Citation slips get corrected against the canonical reference.
His words stay his
Quotes are cleaned but never paraphrased into an editor's voice, and no outside theology is imported. What you read is what was preached.
The same shape every week
Four fixed sections, checked against a schema on every run. That consistency is what makes an archive searchable in year three instead of a pile of one-off summaries.
How it works
Four steps, none of which involve a person.
Detect
Each church's podcast feed is polled through the weekend. The new main Sunday message is identified and everything else is left alone.
Transcribe
The audio is transcribed in full on our own machines — no third party ever receives the recording.
Edit
One pass produces all four sections together, which is how the editor keeps them from repeating one another.
Publish
The note is typeset, filed by date, and delivered — to the web, to an inbox, to the group's channel.