Zortix
Ask Zortix…⌘KSign in
The method

AI stock research, read out of what people actually said

TRANSCRIBED · EXTRACTED · ATTRIBUTED TO THE SPEAKER AND THE MOMENT

Most stock research asks you to trust a conclusion. Zortix does the opposite: it takes the finance podcasts where investors, analysts and operators argue their positions out loud, and turns those conversations into ideas you can check — each one carrying who said it, on which show, and the timestamp where they said it.

The “AI” part is narrow and worth being precise about. Language models do the transcription and the extraction: reading hours of audio and pulling out the distinct claims, the reasoning behind them and the risks the speaker named. They do not form a view. There is no model deciding a name looks attractive, no score, no target price. Every opinion on this site belongs to a person who said it on the record.

How the research is produced
01
Transcribe

Finance podcast episodes are transcribed with timestamps, so every sentence stays anchored to the second it was spoken.

02
Extract

A language model reads the transcript and pulls out each distinct investment idea, with the names discussed, the reasoning given, the stated risks and what would validate it.

03
Attribute

Every idea keeps its source: the show, the episode, the speaker, the moment. Nothing is summarised into an anonymous consensus you cannot check.

The names being discussed right now

Ranked by how much the shows have talked about them, not by anything Zortix thinks of them. Each links to every attributed idea about that name.

Questions

How shows are selected, what counts as a mention, what the depth bands mean, how disagreement is handled, and what Zortix deliberately does not do — answered on the methodology page.

EVERY IDEA IS SOMEONE ELSE'S, ATTRIBUTED AND LINKED TO ITS SOURCE · NOT A RATING, FORECAST, OR RECOMMENDATION