# Molecule Notes vs Logseq

> Write on today's page, link as you go, and every mention gathers on the page you linked. Invite people to write alongside you, and export any page as Markdown.

**A journal outliner you can write in together.**

## Logseq is now two products

> "Every feature, bug fix, and UX improvement must be built twice, resulting in slower development." — [Logseq: Big update: Logseq is splitting into two versions](https://logseq.io/page/b2ad9ce1-9cb7-4436-8083-54cb4516d324/df4dc09d-0a12-4c87-904e-22a9bf4c350a)

On April 24, 2026, Logseq split into two products: Logseq OG, for file-based Markdown graphs, and Logseq, for database graphs. The reason Logseq gave is an honest one, and anyone who has kept two data layers in step will recognize it. It was a defensible engineering decision.

It also changed the thing many people chose Logseq for: notes as plain Markdown files on their own disk. If that's why you chose it, the split may have you weighing where your notes go next.

Be clear about one thing before you look at Molecule Notes: it doesn't store files on disk either. Notes are bullets in a database, and any page exports as Markdown. It's a web app, and it needs a connection. If files on your own disk are what matter most, Logseq OG keeps them.

What Molecule Notes keeps is the part of Logseq that was about writing. On a journal page every note is dated, so nothing needs filing. Link a person, a book or a project with [[double brackets]], and every mention of it gathers on its page under Linked mentions. The graph view shows the connections you've made. And the notes can have other people in them.

- **The current version stores a database.** Markdown Mirror, opt-in per graph, writes a plain Markdown projection of your blocks to disk. In Logseq's words, "The two-way version — where edits in the markdown file flow back into the DB — is in active development on a feature branch and not yet shipped."
- **Markdown export can't hold everything.** Standard Markdown export "cannot capture all data in a graph." EDN is "the only export type that fully captures a graph's data and is editable."
- **It's in beta.** "The DB version is in beta status while the new mobile app and RTC is in alpha. This means that data loss is possible…" Logseq recommends backups and a dedicated test graph.
- **Logseq OG gets maintenance, not new features.** It continues to receive security fixes and Electron and dependency upgrades. "Our focus will be on maintenance and reliability rather than new feature development…"

## Molecule Notes and Logseq, side by side

| Feature | Molecule Notes | Logseq |
| --- | --- | --- |
| Notes as Markdown files on disk | No: Stored as bullets in a database. Any page exports as Markdown. | Partly: Logseq OG keeps file-based Markdown graphs. The current version stores a database, with an optional read-only Markdown mirror. |
| Markdown export | Partly: Any page, one at a time, with Copy as Markdown or Export as Markdown. | Partly: Standard Markdown export can't capture all of a graph's data. EDN is the full export. |
| Journal pages and [[links]] | Yes: Every day has a page. Every mention gathers under Linked mentions. | Yes: Journals, [[links]] and Linked References. |
| Real-time collaboration | Yes: Invite people to a workspace and see each other's cursors as you write. | Partly: In alpha. A paid feature, currently invite-only, open to active monthly backers on Open Collective. |
| On a phone | Partly: No native app. The web app installs to your home screen. | Partly: For the current version, the iOS app is invite-only and the Android app hasn't opened for alpha testing. |
| Plugins | No: No plugin directory. Extensions are workspace agents you @mention on a bullet. | Partly: 65+ plugins support the current version's database graphs. |
| AI in the app | Yes: Name an agent on a line. It writes what it found under your line. | Not documented: None documented. An optional MCP server lets AI applications connect to a graph. |
| Open source and self-hosting | No: Neither. | Yes: AGPL-3.0. Sync and Publish can also be self-hosted. |
| End-to-end encryption | No: Not offered. | Yes: Sync encrypts your graph on your device, with a password you set. |

## Writing in the same notes as someone else

Logseq's real-time collaboration is in alpha: a paid feature, currently invite-only, and open to active monthly backers on Open Collective.

In Molecule Notes, collaboration is a workspace you invite people to by email, each with a role: owner, admin, member or viewer, where a viewer can read and not edit. There's no sharing page by page. Everyone writes in one place, on the day's page, linking what each note is about, and it sorts itself out later on the pages those links point at.

You see each other's cursors as you type, and comments sit on the bullet they're about, where they can be resolved.

## Rebuilding a workflow, either way

Logseq's docs count 65+ plugins that support database graphs. If your setup leans on plugins that don't, moving to the database version means rebuilding some of it. That's worth knowing before you choose where to go.

Molecule Notes has no plugin directory, so it won't replace a plugin one for one. What it has is different. An agent is a colleague you call by name on a line: it reads your pages and the web, and writes what it found under your line, leaving the rest of the page to you. Extensions are workspace agents you @mention the same way.

And like Logseq, it has an MCP server, so your own AI assistant can read and update your notes. A coding assistant can log the day's work on today's journal page, for example. Pages reach the assistant as Markdown with your [[links]] still in them, so it can follow one page to the next the way you would.

## Getting your notes out of Molecule Notes

Any page exports as Markdown, with Copy as Markdown or Export as Markdown. It's one page at a time; there's no button that exports a whole workspace.

To take everything at once, connect an AI assistant to Molecule's MCP server. It can list your pages and read each one as Markdown, with its bullets and their nesting intact.

## Bringing a Logseq graph in

There's no import screen. Import goes through the MCP server: connect an AI assistant to Molecule Notes and point it at your graph's Markdown files, from Logseq OG or from the current version's Markdown export.

It writes pages in bulk. Headings, lists and nesting become bullets, and [[links]] point at pages by title, so Linked mentions build as the pages arrive. A page titled with a calendar day, like 2026-09-13, lands on that day's journal page, so have the assistant title each journal file with its date.

Block references, page properties and task states aren't documented to carry over, so check those pages once they're in. Running the same import twice appends everything a second time rather than replacing it.

## Sources

- [Logseq: Big update: Logseq is splitting into two versions](https://logseq.io/page/b2ad9ce1-9cb7-4436-8083-54cb4516d324/df4dc09d-0a12-4c87-904e-22a9bf4c350a)
- [Logseq on GitHub: license, beta status and data-loss warning](https://github.com/logseq/logseq)
- [Logseq docs: the DB version](https://github.com/logseq/docs/blob/master/db-version.md)
- [Logseq forum: What's new with Logseq DB, May 16th 2026](https://discuss.logseq.com/t/whats-new-with-logseq-db-may-16th-2026/35020)
- [Logseq on Open Collective](https://opencollective.com/logseq)

Canonical: https://moleculenote.com/vs/logseq

Comparison verified September 2026.