Logseq alternatives after the database migration

Logseq is now two products: the original, where your notes are markdown files in a folder, and a database version that keeps the graph in SQLite. If you picked Logseq for plain files, the difference between them is worth understanding before you decide whether to stay or move.

Logseq built its reputation on the same premise as Obsidian: notes as plain markdown files in a folder you control, with an outliner and a graph view on top. That premise is no longer the whole story. Logseq now documents a second product alongside the original: a database version that keeps your graph in SQLite rather than in a folder of markdown files. The two differ in exactly the way that matters if file ownership is why you chose Logseq in the first place.

What actually changed

Nothing in Logseq's published documentation describes an automatic conversion. Moving a file graph across is something you go and do: the DB Graph Importer sits behind the Import menu, and in the documentation's own words it "converts a file graph to a DB graph" by importing your markdown files. So an existing file-based graph is not going to change shape under you while you are not looking. What has changed is which of the two products the database documentation is written about, and that matters more than it first appears, because it is the one that stores your notes somewhere you cannot open directly.

Why the split happened at all

This is a reading rather than a statement from the team, and it is worth more than assuming a monetization motive. Maintaining two fundamentally different storage architectures inside one codebase, a flat-file backend and a database backend, means every feature has to be built twice, tested twice, and explained twice to users who do not know which mode their graph is running in. That is a real engineering cost, and it is the same tension that appears whenever a tool tries to serve both the plain-text purists and the people who want fast queries across tens of thousands of blocks. Two products let each side move at its own pace instead of both being held back by the other.

What the database version gets you, and what it costs

Start with what it is not. This is not a move to the cloud in the way Notion or Roam are cloud-hosted. Logseq's documentation has the database written to your own machine, at ~/logseq/graphs/<graph>/db.sqlite. Your notes stay on your disk. What you give up is not privacy, it is direct access: the file you own is now a database file rather than something you can open in a text editor and read.

The export options are where that trade shows most clearly, and Logseq is refreshingly plain about it. There is an "Export as standard Markdown (no block properties)" option, and the documentation says of it that "it cannot capture all data in a graph." It points instead at an EDN export, described as "the only export type that fully captures a graph's data and is editable," while adding in the same breath that the EDN export "is not yet recommended as the only means to backup a graph." Put those two sentences beside each other and the position is clear enough: the readable export is lossy, and the complete one is not yet something the team wants standing alone as your backup.

What to check before you commit either way

If you're already running Logseq and unsure which track you're on, check which app you actually downloaded and whether your graph has a folder full of .md files sitting next to it on disk. If it does, you're almost certainly still file-based, and nothing about your day-to-day workflow changes on its own. If you're starting fresh and deciding whether to adopt Logseq at all, the honest framing is that you are choosing which trade you want: a database version whose capability is what Logseq now documents in depth, with your notes in a file you cannot read directly, or the original file-based version, where the notes stay yours in the literal sense and you are betting on plain text rather than on a roadmap.

Where people are actually looking

Two different audiences read a page like this, and they want different things. If you want to keep the plain-file promise and never touch the database version, Obsidian vs Notion AI is the closest comparable: Obsidian never had a Logseq-style split, and its files-in-a-folder model is unlikely to change under you. If you're specifically after Logseq's outliner-plus-linking style, its Zettelkasten with AI style of atomic, linked notes maps closely to what Logseq's block references were built for, and works in Obsidian or plain files just as well.

If what actually drew you to Logseq was the promise of AI eventually built in, rather than the outliner itself, it's worth reading the AI note apps comparison before committing to either Logseq track, because most AI-native apps make the same ownership trade the database version is edging toward, just further along.

Whichever way you land, the deeper question is the one covered across this whole note-taking app comparison: not which outliner has the nicest graph view, but which one still lets you read your own notes without an app's cooperation five years from now.