It reads all of them, not a sample
The whole project goes into the question: every transcript plus the same totals the insights tab computes. No search box to guess the right keyword for.
You don't need to watch all of it. You need to know which track opens the album, and who said the opposite.
The whole project goes into the question: every transcript plus the same totals the insights tab computes. No search box to guess the right keyword for.
Answers come back with the responses behind them, so you can jump straight to the twelve seconds where someone said it.
An answer that came in a minute ago is already counted. Ban or delete a response and it disappears from the next reply — there's no index to rebuild.
Names never leave the building: the model sees response identifiers, and the names get resolved in your browser when the citations render. If a project is too large for one prompt, the newest answers win and you're told plainly how many of how many were read.
A chat is yours. It's private inside your organisation rather than shared across everyone with access to the project.
It is a chat over every response in a project at once. You ask in plain language, and the answer is built from the transcripts, with the responses it leaned on quoted underneath so you can jump to the exact seconds.
Every one. The whole project goes into the question: all transcripts plus the totals the insights tab computes. If a project is too large for one prompt, the newest answers win and you are told how many of how many were read.
No. The model sees response identifiers only. Names are resolved in your browser when the citations render.
No. A chat is private inside your organisation to the person who started it, rather than shared with everyone who has access to the project.
The use cases built on top of it. Each page walks the workflow end to end.
Give them a way to say it. We'll walk through what Wall of Voice could look like for your audience.
See what Wall of Voice could look like for your audience.







