Two channels, covered all the way down to the full text.
Plenty of tools already do those things well. We do the two channels they can't reach.
We transcribe episodes ourselves when the publisher provides no transcript, so the whole conversation is searchable text, not the title and the show notes. Speakers are labelled where the audio allows it.
The body of the issue, not the summary. Newsletters without a public feed reach us as email, which we parse into the same structure as everything else.
Phrase and boolean queries across every transcript and issue we hold, ranked by relevance or date, with the matching snippet returned alongside each hit.
An alert is a query that keeps running. New matches are delivered by email, Slack or webhook, usually within minutes of publication, with a snippet and the source attached.
Connect Claude, ChatGPT or any MCP client and it can search, read posts, manage collections and set up alerts directly. OAuth, no key pasting.
The same data as JSON: search, sources, posts, people, collections, alerts and alert hits. Cursor paging, documented fields, one token.
We pull the paid placements out of every episode and issue: the ad reads, the "brought to you by", the presenting sponsor. So you can see who is advertising on which shows, and how often.
Every source carries an authority score, a publisher type and audience signals. Hosts and guests are resolved to people, so "every episode this exec appeared on" is one query.
Scheduled summaries delivered where you already work. Every claim carries its source, date, quote and audience size, so the output is forwardable to a client as-is.
Same data, different question. Things customers actually set up.
14 days free, nothing charged until the trial ends, cancel any time.