01
Paste a transcript or link
Any recording that has a transcript is a source. Paste the text, drop a link, or upload a caption file. Add a short voice sample, a few paragraphs of how you write or talk, and Verso is ready to run.
How it works
What happens between paste and output
Three steps on the surface. Underneath, five, and the one worth understanding is the check that discards any moment we cannot locate in your own transcript.
The flow
Three steps
02
Verso finds the strongest moments
The whole transcript goes through in one pass. Verso reads it like an editor looking for the clip: the story with the most tension, the line that would stop a scroll, the argument that earns a share. Each moment is scored against the rest.
03
You get a board, ready to copy
Clip scripts with hooks and captions, an X thread, shorts captions, a LinkedIn post, and a newsletter issue, all in your voice. Edit anything, regenerate any format, copy with one click.
In detail
The five stages
Written out because the trustworthy part of this product is not the model call. It is what happens before and after it: what goes in, what gets checked, and what gets thrown away.
No fabrication
Every moment Verso surfaces is quoted verbatim from your transcript and verified against it before you see it.
- 01Ingestion
Getting the words out
A pasted transcript is already text. A link or caption file is not: Verso pulls the transcript from it server-side, normalises the spacing, and checks that what comes back is readable text. If the link has no available transcript, Verso tells you rather than running on an empty string and producing confident output about nothing.
- 02Classification
What kind of recording is this
Before finding moments, Verso reads a sample of the transcript to work out the format: a solo episode, an interview, a keynote, a panel. That determines which output mix to produce and how to read the moments. A conversation has different peaks than a monologue. This step is fast and cheap and runs first.
- 03Moment detection
Reading the whole thing
The full transcript goes to the model in one pass where it fits. Finding the best clip in an hour of audio is not keyword search: it takes reading context. A line is only a great hook if the minute before it set something up. Long recordings escalate to a frontier model, and anything over the size limit is reported, not dropped silently.
- 04Verification
Every moment checked against your text
The model returns a quote for each moment it selects. Before you see it, that quote is located in your own transcript, first exactly, then allowing for whitespace differences, then by the longest matching run that does appear. A quote that cannot be found is dropped. A highlight pointing at the wrong sentence is worse than no highlight.
- 05Rewriting
Each format written differently
A clip script and a LinkedIn post start from the same moment but are not the same piece of writing. A hook has to work in the first second. A thread needs a narrative arc across its posts. A newsletter has room to expand. Each format is written for how it is read, not resized from a single block of text.
What it looks like
The same moment, five formats
These are the kinds of outputs a creator gets from a single podcast moment. The source: an episode about treating a recording as a content mine rather than a finished product.
Clip script
Hook
Stop treating your podcast like a finished product. It is a source.
Caption
Every hour you record has forty good clips in it. You cannot see them from the outside. The people winning on short-form right now are mining the episode, not just posting it.
Thread opener
"You recorded for an hour. You posted once. Here is what you left on the table."
Thread continues across 6 posts with the argument from the episode.
Newsletter lede
Subject: The episode is the mine
"Ran an experiment last week: took one ninety-minute episode and treated the transcript like a source document. What came out of it surprised me."
All of the above are rewritten in the creator’s voice, not in a neutral template style. The voice sample you provide determines how every output sounds.
Questions
Before you paste anything
See it on your own recording
One project a month is free. Paste a transcript and watch what comes out.