SumizAI vs Logseq: bullets you write against answers that file themselves

Logseq is a free, open-source outliner: every page is a tree of bullet blocks, each day opens with an empty journal page, and the graph lives on your own disk as Markdown or Org-mode files.

The two apps share one belief, that your notes belong in local files, and almost nothing else. In Logseq you type every block. Here you talk to a model, and the replies you want to keep become AI notes with a title, a summary and the full exchange kept as the source. Logseq facts below come from logseq.com and its GitHub page, read on 23 October 2026.

Table of contents of a vault, with numbered chapters filled with AI notes by the model
No daily page here: the vault opens on a table of contents with numbered chapters.

Logseq next to an app for AI notes

Logseq and SumizAI side by side, as of 23 October 2026
SumizAILogseq
Author of each noteThe AI model, working from your conversationYou, block by block
PriceSeven days free, then 1 USD a month; model usage billed by your providerFree forever for personal use
LicenceProprietary subscriptionOpen source, AGPL-3.0
PlatformsMac version coming to the Mac App Store; Windows, iOS, Android latermacOS, Windows, Linux, iOS, Android, web app
FilesOne .md per note; a local database indexes themMarkdown or Org-mode files; the newer DB version (beta) keeps graphs in SQLite
Shape of a noteTitle, one-line summary, body, original question and answerAn outline of nested blocks
OrganisationChapters numbered 1., 1.1, a), a.1)Journal pages, page links, block references, queries
ExtrasDuplicate check before saving, PDF and PowerPoint exportFlashcards, PDF annotation, task management, 150+ plugins, 30+ themes
AISeven providers on your own API key, Ollama includedNot among the listed features; plugins extend the app
SyncNoneLogseq Sync in beta; RTC sync and collaboration in alpha

A journal page against a chapter tree

Logseq starts every day the same way: today's date at the top, a cursor on the first bullet. You write met Anna about the grant, indent, add a task, link a page. Structure grows from the bottom up, through links and references, and a query can later pull all open tasks tagged with a project onto one page. It rewards people who write a lot, every day.

This app has no daily page at all. Picture a PhD student on a Thursday asking Claude whether a Mann-Whitney test suits two small samples. The reply turns into one of her AI notes, filed under 2.1 Methods, sub-chapter a) Statistical tests, because the model read the vault's table of contents and picked the closest number. If nothing fits, it asks for a new sub-chapter.

By Friday the Logseq journal shows what happened on Thursday. The vault of AI notes shows what you learned, sorted by subject, with the date as a detail.

Both are valid. They answer different questions: “what did I do?” against “what do I know?”

Moving between a Logseq graph and a vault

From Logseq. A file-based graph is already a folder of Markdown files, so there is nothing to export. In the import panel pick that folder, or a few pages. The importer takes .md, .markdown, .txt and .pdf, one file at a time, and writes each into a new AI note with the file's original text attached as the source. Org-mode .org files are not on that list, and graphs kept in the DB version need exporting to Markdown first.

Expect the outline to change. Nested bullets become prose with a summary on top; block references and query syntax survive only inside the source text.

To Logseq. Download the vault as a ZIP. Inside sit base.md with the table of contents and the folder notes/. They are ordinary Markdown files, readable by any tool that reads .md, a Logseq file graph included.

Stay with Logseq if this sounds like you

  • You think in outlines and like to fold, drag and re-indent blocks.
  • The daily journal is your inbox, and tasks with deadlines live next to your notes.
  • You study with spaced repetition: Logseq has flashcards built in.
  • You read papers as PDFs and annotate them where the notes are.
  • You use Linux, or want the same notes on an Android phone today.
  • You want open source and zero cost. Logseq is AGPL-3.0 and free for personal use, while this app is closed and paid.

A Logseq power user with 30 queries and a flashcard deck loses a lot by switching. Keep it.

Pick the vault of AI notes if this is your week

Your best material comes out of conversations with GPT or Claude, and the journal page stays empty because copying an answer into bullets feels like homework. That is the one case this AI notes app was made for.

You also get things Logseq leaves to you. Before saving, the app looks for a near-duplicate with full-text and trigram search, a second model scores the match from 0 to 1, and you decide whether to merge. Merging only appends; the original source is never overwritten. Ask a follow-up a month later and the model receives the table of contents plus up to five relevant notes, within 24,000 characters.

And it hands work over in formats people open: a PDF with summaries only or with full AI notes, or an editable .pptx deck in table-of-contents order.

Plenty of Logseq users try Obsidian too, for the same local files. If that is where you are heading, read how this Obsidian alternative fits next to it: vaults exported from here open in Obsidian as they are, links included.

Logseq, Obsidian alternatives and AI notes: questions

Does SumizAI use wiki-links like Logseq?

No. The model writes ordinary Markdown links such as [Mann-Whitney test](mann-whitney-test.md), and only to notes that exist. Every note lists its backlinks underneath.

Can I keep a daily journal among my AI notes?

Not in the Logseq sense. There is no date page; notes go into chapters by topic. If your system is the journal, stay where you are.

Is my data local in both apps?

Yes. Logseq stores the graph on your disk, and so does this app: plain files plus an index, with no server holding your notes. Requests go from your Mac straight to the AI provider you chose, after you allow them.

What does a switch from Logseq cost?

1 USD a month after the 7-day trial, plus one model call per imported file on your own key. A local Ollama model makes the import free of API fees.

Is it one of the Obsidian alternatives worth trying after Logseq?

Only if your notes start in AI chats. For a hand-written outline, Logseq and Obsidian both do more.

Logseq features and licence terms: logseq.com, platforms: logseq.com/downloads, licence, file formats and DB version status: github.com/logseq/logseq. Pages opened on 23 October 2026. The price of Logseq Sync is not published there, so it is left out. Everything said about this app is checked against its source code and its App Store listing.

Other comparisons

Let the answers file themselves

Try it free for seven days, next to Logseq if you like. Ask one real question a day and look at the chapters on Friday.