Skip to content
How to start Logseq without building a second brain
How-To

How to start Logseq without building a second brain

Sam RiveraNotes, writing & local-first

What you’ll get

A Logseq graph you can open today: today’s journal takes the notes, pages appear when topics earn them, and nothing depends on a plugin shopping list. 30-45 minutes if you stop when this guide stops. No query language. No “second brain” architecture diagram.

Logseq is not Obsidian with different icons. It is a lab notebook. You write in bullets on a date, then link out. If you try to build a folder tree on day one, you will hate it for the wrong reasons. For the fork between stacks, see Obsidian vs Logseq. If you already know you want document files in folders, use the minimal Obsidian start instead.

Before you start

  • Logseq installed from logseq.com (desktop app; free and open source)
  • A folder you control for the graph data: something like Documents/logseq-graph, not a temp download folder
  • Optional later: sync (git, Syncthing, or Logseq Sync). Local-first for week one is fine

Mental model before you click anything: almost everything starts as a bullet on today’s journal. Pages are destinations you create with [[double brackets]] when a topic keeps showing up. You are not naming files and choosing folders first. That decision delay is the thing Logseq removes on purpose.

Skip this guide if you need long-form essays in one continuous document as the default. Logseq can hold longer writing, but the happy path is outlines and journals. Document people usually end up happier in Obsidian.

Steps

1. Create a local graph in a plain folder

  1. Open Logseq.
  2. Choose Add a graph (or equivalent first-run prompt) and pick Create a new graph.
  3. Point it at an empty folder you own: Documents/logseq-notes is fine. Name the graph something boring like Notes.

Logseq will create its own structure inside that folder (journals, pages, config). You do not need to pre-build PARA folders. Do not put the graph inside a half-synced cloud folder until you understand conflict copies. Desktop-only for a week is the cheap way to learn.

Open the left sidebar and confirm you land on Journals. That home screen is the product. If you immediately hunt for a file tree like Finder, pause. The tree is secondary here.

2. Write everything on today’s journal

Click into today’s journal page. Type a bullet. Hit Enter for the next one. That is the whole capture loop for week one.

Dump real stuff, not demo fluff:

  • Meeting notes under a parent bullet with the meeting name
  • Random ideas as single lines
  • Tasks with TODO at the start of a block (Logseq turns those into checkable tasks)
  • Links you want later as plain URLs or [[Read later]] children

Do not create a page for every fleeting thought. Official Logseq guidance is blunt about this: most beginners should write on the journal first. The date is the filing cabinet. “Where does this go?” is answered by “today,” which is good enough until a topic earns its own page.

If you catch yourself staring at an empty [[New Page]] wondering what to name it, cancel and write the bullet on the journal instead. Naming is organization work. Capture is thinking work. Do the second one first.

When the same subject shows up a second or third time, wrap it in brackets while you type:

Talked to Maya about [[Home server backups]]

Click the link. Logseq creates (or opens) that page. Put one sentence at the top that says what the page is for. Below that, nest bullets that belong to the topic forever, not just today.

Rule of thumb for the first two weeks:

  • Once: leave it on the journal
  • Twice: consider a page link next time you mention it
  • Three times: it deserves a page, and you will find it via linked references

Hashtags like #idea work too. Treat them as light filters, not a taxonomy project. Five tags max until you feel actual pain finding things. Empty pages with pretty titles and no content are PKM furniture. Delete them without guilt.

4. Indent like an outline, not like folders

Logseq’s unit of thought is the block (one bullet). Indent under a parent to keep detail attached without creating a new page.

Example on today’s journal:

- Project sync call
 - Goal: decide backup cadence
 - Decision: nightly to [[Home server backups]]
 - Follow-up TODO email Maya the cron line

Tab and Shift-Tab move indentation. Collapse a parent bullet when the meeting is over so the journal stays scannable. This is where outliners win: context stays nested under the event instead of scattered across five files named after moods.

You are not building a directory tree. Indentation is structure for this thought, not a permanent taxonomy of your life. If a nest gets deeper than three levels often, you are probably over-outlining. Flatten and move on.

5. Add a tiny daily template (optional)

If a blank journal still feels too open, give yourself three headings and stop. Create a page called templates. Under a parent bullet, add:

- Daily
 template:: Daily
 - Log
 - Scratch
 - Tasks
 - TODO

The template:: Daily line is what makes that block insertable. On today’s journal, type /template, pick Daily, and press Enter.

Do not build a mood tracker, habit grid, or weather widget into this. Three sections is enough: what happened, messy thinking, and tasks that appeared. You can wire config.edn later so new journals auto-load the template. That is a week-two chore, not a day-one requirement. A template you insert by habit beats an automation you never trust.

6. One weekly ritual (10 minutes)

Once a week, open Journals and skim the last seven days. You are hunting for repeated topics and orphan bullets that still matter.

  1. For each repeating topic, make sure it has a [[Page]] and that today’s or this week’s mentions link to it.
  2. Open each active page and check Linked references. Promote anything that should live on the page permanently by copying or moving the block up.
  3. Mark finished TODO items done. Delete noise. Do not “organize” for sport.

No graph-view meditation. No tag rename marathon. Ten minutes of promoting what already earned attention. If nothing repeated, your journal did its job and you leave the graph alone.

7. Sync when you need the phone, not before

Use Logseq on one desktop for a week. When phone capture becomes the actual pain, pick a sync path:

Method Tradeoff
Logseq Sync Official, least fiddly if you want their cloud
Git Free, great for text nerds; phone story is weaker
Syncthing Free, you own the pipes; read conflict docs
iCloud / OneDrive / Dropbox Tempting; conflict copies will eventually teach you humility

One sync method. Never two. Migrating an almost-empty graph is cheap. Migrating a conflicted mess is a weekend you will resent. Same advice as the Obsidian minimal guide, because sync physics do not care which app wrote the Markdown.

Common mistakes

  • Creating a page for every bullet on day one, then wondering where anything is
  • Installing a marketplace of plugins before you have two weeks of journals
  • Treating the graph view as proof you are productive
  • Building a folder hierarchy that fights the outliner (Logseq is not PARA cosplay by default)
  • Syncing through two cloud services at once and collecting conflicted copy files like trading cards
  • Trying to write novel-length essays as one unbroken page when you actually think in outline nests

Tooling that helps

Wrap-up

You have a local graph, a journal habit, and a rule for when pages earn their keep. That is enough Logseq to think with. Add queries and plugins when a specific retrieval job hurts, never because a YouTube thumbnail promised a second brain.

Who should skip this: if you need real-time co-editing with non-nerds, stay in Notion or Docs. Logseq shines when the bullets are yours and the date is enough filing for most of the week.

Related