Podcast to Obsidian or Logseq: A Second Brain That Actually Gets Fed

A second brain has one job: receive things and make them findable later. Most podcast setups fail at the first half. You listen in one app, think "I should save this", and the thought dies between the player and the vault.

The fix is less about plugins than about deciding what an episode note is. Is it a copy of the summary, a page of quotes, or three sentences with links? Pick one, do it the same way every time, and the vault starts to compound.

Obsidian and Logseq, brieflyCopy Anchor Link

Both read Markdown; the difference is the shape.

Obsidian is file-first. A note is a file, frontmatter is metadata, links are [[wikilinks]]. It suits "one episode, one note" with properties you can query.

Logseq is block-first. Everything is an outliner block, properties attach to blocks, and the daily journal is the default inbox. It suits "one idea, one block" with the episode as a property.

Pick based on how you think, not on a feature list. If you write paragraphs, Obsidian. If you think in bullets and outliners, Logseq.

The note that works in bothCopy Anchor Link

Whichever app you use, an episode note needs four things: the claim, the supporting detail, the source with a timestamp, and a link to a topic.

---
show: "The Show"
guest: "Guest Name"
date: 2026-09-14
url: https://app.podwise.ai/dashboard/episodes/1234567
topics: [pricing, hiring]
---

## Switching cost beats feature velocity

The guest's argument: features are easy to demo, switching costs are
invisible and do the actual work.

> "Nobody buys a migration. They buy not having to migrate." — 00:41:12

My take: our export is too easy. Fix that before the roadmap.

Two things make this useful later. The url plus timestamp lets you verify the quote in one click. The topics property gives the graph something to link to.

Getting episodes inCopy Anchor Link

Export from Podwise. With Obsidian or Logseq connected in Settings › Connections, an episode exports as Markdown formatted for that app: summary, chapters, highlights, and the transcript if you ask for it. The Obsidian and Logseq pages show what the files look like.

Let an agent write the note. The MCP server's export_episode_markdown does the same thing from a chat: "export this episode to Obsidian with five highlights" produces the file you would have written by hand. Good for the episodes you only half-listened to.

Write it yourself. If the episode gave you one idea, skip the export. Open the transcript, copy the quote, add your line. A vault of 50 real notes beats a vault of 500 imported summaries.

The transcript questionCopy Anchor Link

Full transcripts are wonderful for search and terrible for a graph. They swamp the note, break link previews, and make the file 40 times larger than the idea inside it.

The version that works for most people:

• Keep the summary and highlights in the episode note.
• Keep the transcript in a separate folder (transcripts/) only if you actually search it, or skip it — the timestamp link opens the transcript in Podwise anyway.
• Never let the transcript be the note. If future-you opens the file, the first screen should be the idea.

If you do keep transcripts, name them by episode and link to the note, not the other way around. That way the graph shows ideas, not raw text.

Making the graph do somethingCopy Anchor Link

Once the properties are consistent, the vault starts answering questions:

• By topic: a Dataview query over topics produces a reading list per subject, built from what you actually listened to.
• By show: a page per podcast, with its episodes backlinked. Useful when a host repeats an argument across guests.
• By claim: the notes with a my-take section but no outgoing links are the ones you have not processed. Those are the ones to revisit.

In Logseq the same idea is a query over page properties, and the daily journal becomes the inbox: capture a block during the day, tag it with the episode URL, move it to a topic page on Sunday.

The maintenance nobody warns you aboutCopy Anchor Link

Two rules keep this from becoming another graveyard:

1. A note with no topic link is unfinished. Either add the link or delete the note.
2. One episode can produce more than one note. If a conversation gave you two unrelated ideas, split them. The vault links ideas; the episode is metadata.

Do that for a month and you will notice something: the podcast notes are not the valuable part. The links between them are. That is the second brain doing its job.