Jotura vs Logseq: outliner blocks or plain markdown documents

Logseq is an outliner and Jotura is a document editor. In Logseq every line you write is a bullet, every bullet is a block with its own identity, and you can reference or embed any block anywhere in the graph. In Jotura you write pages of ordinary markdown, and saving an untouched file leaves the bytes on disk unchanged. That one difference decides most of this comparison.

Stay on Logseq if block references and the graph view are how you think. Choose Jotura if you want files any other program can read and a fast editor for long documents. It also ships a command line tool that lets AI coding agents edit your notes safely. Both apps keep your data on your own machine, and both speak markdown, though Logseq 2.0 changes what that means.

Blocks versus documents

In Logseq, structure lives inside the app. A block reference like ((6512a...)) means something to Logseq and nothing to a text editor. That is a fair trade: you write one idea once and pull it into as many contexts as you need.

In Jotura, structure lives in the text. Headings are headings, lists are lists, and a link is a [[wikilink]] that any markdown tool understands. The cost is that you cannot pull a single bullet into six other notes.

Where Logseq wins

Logseq does several things Jotura does not do at all.

Logseq has	Jotura has
Block outlining and block references	Document editing only
A graph view of your whole vault	No graph view
Plugins and community themes	No plugins
Flashcards with spaced repetition	No flashcards
Whiteboards in Logseq OG	No canvas or whiteboard
A query language based on Datalog	No query language
Zotero integration	No reference-manager integration
Publishing a graph as a public website	No publish to web
Open source, AGPL licensed	Closed source
iOS and Android apps	Android only, no iOS
PDF annotation inside the app	PDF ingestion for search, not annotation

Logseq is also free in a way Jotura is not. The app costs nothing and its source is public, so if the project stopped tomorrow you could still build it yourself. Jotura is closed source, and its sync service costs money. No feature list cancels that out.

Daily notes, backlinks, and checklist tasks exist in both apps, so none of them is a reason to switch. Logseq’s journals and linked references are core to how it works, and its TODO and LATER system collects tasks at least as well as Jotura’s vault-wide task list.

Where Jotura wins

Jotura keeps a byte-identical round trip. Open a file, save it without editing, and the bytes are unchanged, down to your original list markers and spacing. Logseq OG rewrites files in its own outline dialect, adding property lines such as id:: to blocks. If git, a script, or another editor touches the same folder, that difference shows up in every diff.

Jotura’s editor shows formatted text as you type, what you see is what you get, with a toggle into raw markdown. Logseq gives you one bullet per line, which is excellent for thinking and awkward for a two-page document. Jotura also has a real table editor instead of pipe characters you align by hand.

Search is where the Rust core shows. Keyword search across 10,000 notes takes about 8 milliseconds in a release build. Logseq’s search is quick as well, and its query language does things Jotura cannot.

Semantic search is the clearer gap. Turn it on in desktop Settings and Jotura downloads an embedding model once, then finds notes by meaning with no API key and nothing leaving your machine. Logseq has no built-in equivalent. Document ingestion turns PDFs, Word files, and spreadsheets into searchable markdown mirrors, so a contract shows up in the same search as your notes.

The file question, and Logseq 2.0

For years the strongest argument for Logseq was that a graph was just a folder of markdown files. That changed on 13 July 2026 with the 2.0 beta, which moves the canonical store into a local SQLite database. Markdown becomes an export rather than the live source. The file-based app continues as Logseq OG, maintained with fixes but no new features.

That is the state as of August 2026, and worth reading before you plan your next few years. Jotura’s answer has not changed: the folder of .md files is the product, and the search index is disposable.

What moving from Logseq actually looks like

Jotura opens an existing Obsidian vault as it stands. A Logseq graph is a different story.

Point Jotura at a Logseq OG folder and the files load, because they are markdown. They will not read as documents. Nearly every line starts with - , so you get a wall of list items. Property lines like id:: and collapsed:: true appear as plain text, and so do :LOGBOOK: drawers.

Block references are the real loss. A ((uuid)) resolves only inside Logseq, so in Jotura it sits there as inert text that nothing else can follow. Headings, wikilinks, and checklists carry over fine.

On the 2.0 beta, export standard markdown first, because a SQLite database is not a folder any editor can open. Either way, copy the graph somewhere else and try it there, rather than pointing a new editor at your only copy.

Working with AI agents

Logseq OG’s files are plain enough that an agent with shell access can edit them today, and Logseq’s plugin API is a real automation surface Jotura has no answer to. What Logseq does not have is a first-party command line tool.

Jotura ships one, also called jotura, alongside the app. It came after the editor, not before it. The tool checks a hash before every write, requires a replacement to match exactly once, writes atomically, and leaves frontmatter alone when it edits a note body. The safe loop reads the current hash, then hands it back.

HASH=$(jotura read "Projects/Billing.md" --json | jq -r .hash)
jotura edit "Projects/Billing.md" --replace "Postgres" --with "SQLite" --if-hash "$HASH"

Agents can also record decisions and recall them later, so a choice made in one session is available in the next.

jotura memory log --kind decision --title "Chose SQLite over Postgres for the local cache"
jotura memory recall "database choice"
jotura context "billing rewrite"

Agent configuration lives in an Agents/ folder inside the vault. One command, jotura agents sync, writes it out to each tool’s expected location, so Claude Code, Codex, Gemini, and Copilot all read the same instructions. To wire up the agent skill itself, jotura skill install all covers Claude Code, Codex, and Gemini. The CLI reference and the guide on persistent memory for coding agents go deeper.

Sync and money

Logseq Sync costs $5 a month, or roughly $4.17 a month billed annually. Jotura Sync is £4 a month or £40 a year, with no free trial. Those are close enough that price should not decide this. Plenty of people skip both and sync a folder with Syncthing, Git, iCloud, or Dropbox for nothing.

Jotura Sync is end-to-end encrypted. Keys derive from your passphrase using Argon2id, a deliberately slow key-derivation function. Each file gets its own key, and file paths are hashed, so the server never sees a filename.

Sharing comes with it: encrypted folders or single notes, editor and viewer roles, and automatic merge of concurrent edits. Version history goes back three years with attribution. Both people need Sync for sharing to work. The details are on the sync page and pricing.

Which should you pick

Choose Logseq if you think in bullets and block references. Choose it if the graph view earns its place, if you want plugins or flashcards, or if you need iOS. Choose it if you publish notes to the web, or if open source is a requirement rather than a preference.

Choose Jotura if you write documents rather than outlines. Choose it if you want files that stay readable by everything else you own, or if semantic search on your own hardware is useful. It also fits if you want AI agents editing your notes through a tool built for that. Copy a graph to a spare folder, point Jotura at it from the download page, and read a few notes before committing.