Ask Claude, ChatGPT, or any assistant for a podcast playlist. It arrives in Rovyn as a ready-to-play edition — a goal as the title, the assistant as the byline, each episode introduced by a one-line editorial why.
Rovyn doesn't recommend anything itself. Everything on your shelf is there because you asked for it.
"Build me a two-hour playlist on fusion energy — intermediate level, include a skeptic." Your assistant searches, picks, and explains every choice.
One link, one tap. The playlist lands on your shelf like a typeset magazine issue: the goal as the title, the assistant as the byline, a written why on every episode.
Worth my time, not a good pick, or a note in your own words — passed to your assistant as you wrote it, not summarized and not scored. The next edition is better because you said so.
“Where every other podcast app puts Discover, Rovyn puts an empty shelf, waiting for what you ask for. The assistant is the editor; the app is the catalog, player, and feedback system.”
An editor who can read the request builds for it. Ask for a two-hour commute course on the history of interest rates and that is what lands — sequenced to be listened to in order, not a search result. Ask for a voice that disagrees with the others and the dissent is in the edition, with a line saying why it's there. Say "too introductory" in your own words and the next edition is less introductory, because your assistant read the sentence you wrote rather than a number you tapped.
That works because nothing else is competing for the shelf: no Discover tab, no ads, nothing ranked by what would keep you listening. What you asked for is the only input, so there's no algorithm to argue with — just an editor to correct.
Rovyn is designed so your listening stays between you and the assistant you chose: your notes go to that assistant, not to an ad network. Rovyn speaks Cueback, an open protocol, and the whole backend is designed to be self-hostable.
Read the privacy policy for the specifics.