LANBrain
Back to homeDocumentation

Guide

Best practices

This is the short version of how to grow a brain that gets more useful every month instead of slowly turning into a junk drawer. None of these habits are heavy. The whole rhythm fits in about thirty minutes a week.

The principle behind all of it is one sentence: atomic notes that link. Small notes about one thing, connected to other small notes. Everything else in this file is just how to keep that principle alive over years.

The weekly rhythm

Set a recurring half-hour on your calendar. Pick any day. Friday afternoon is a good fit because most weeks have produced enough material by then.

Three things happen in that half-hour:

  1. Ingest. Open raw/ and look at what has accumulated. Voice memos, screenshots, articles you saved, the WhatsApp export from Monday's meeting. Ask your AI chat to ingest whatever is new, and let the AI process it.
  2. Query. Ask the brain something you actually want to know. "What did the team agree about pricing last quarter?" "Who introduced me to the Berlin contact?" The point is not just the answer, it is to surface notes that should be more connected than they are. When you catch the AI struggling to find an answer, that is a signal the brain needs more links in that area.
  3. Prune. Move two or three files out of raw/ that have already been summarised. Delete the duplicate. Archive the project that is finished. Five minutes of light cleanup per week stops the bloat that kills brains at month nine.

That is the loop. Ingest, query, prune. You will get faster at it after three or four weeks.

The monthly rhythm

Once a month, give the brain an hour. Three things, in order:

  1. Walk the graph in Obsidian. Open the graph view. Look at the shape of the whole thing. The clusters are healthy, the islands are not. If you see an orphan node sitting on its own, click into it and add two or three wikilinks to the rest of the brain. If you see two clusters that should be connected (your work network and your friend network share three people, say), add the links that make that visible.
  2. Run the linter. Open your AI chat and ask it to run the lint-check skill. It will list broken wikilinks, missing frontmatter, files that look like duplicates, and anything in raw/ that has been sitting there for more than thirty days. Fix what feels worth fixing.
  3. Enrich. Pick the three most important notes from the month (your top project, the deal that moved, the person you met who matters). Ask the AI to enrich them. "Pull in everything you know about this and make the page denser." This is where the brain stops being a filing cabinet and starts being a thinking partner.

The atomic-notes principle, with an example

You had a one-on-one with your boss on Tuesday. She told you the team is going to reorganise in Q3 and your role might shift toward strategy. You also discussed the new hire, the contractor renewal, and a side project she wants to revive.

The wrong way to capture this is one note titled "1-1 with Sarah 2026-05-12" that contains four paragraphs about four different things. It is fine for a week. After a month you cannot find the reorganisation context because it is buried inside the meeting note.

The right way is five small notes:

  • wiki/sources/2026-05-12-1-1-with-sarah.md, the meeting itself, three lines summarising what was discussed, with wikilinks to each topic.
  • wiki/projects/q3-reorganisation.md, a project page that did not exist before, started from this meeting, will grow with more context as the quarter unfolds.
  • wiki/people/new-hire-candidate.md, only if it is a specific person, otherwise a line in the team project page.
  • wiki/decisions/renew-contractor-2026.md, captured as a decision once it is made.
  • wiki/projects/dormant-side-project.md, moved out of archive/ back to wiki/projects/.

Now when you ask the brain "what is happening with the Q3 reorganisation?" the AI lands on a single page that links back to every meeting where it was discussed. That compounds. A meeting-only note does not.

You do not have to do this manually. Paste the raw meeting notes into your AI chat and ask: "split this into atomic notes and file them in the right folders, link them together." The chat will do it. You confirm.

Handling entity disambiguation

The same person shows up three different ways across the brain: "Jane", "Jane D.", "Jane Doe-Smith from Acme". The fix is to pick one canonical page (wiki/people/jane-doe.md) and add the other spellings to its aliases field:

aliases: [Jane, Jane D., Jane Doe-Smith]

Then ask the AI to walk through the brain and re-link every mention of any alias to the canonical page. The chat does it in one pass. You review.

The same applies to organisations and concepts. One canonical page per real-world entity. Everything else points to it.

Archive versus delete

If something is finished but interesting, move it to archive/. Old projects, decisions that were taken and executed, deals that closed. The archive is searchable, the AI still reads it for context, but it does not clutter your active wiki.

If something is wrong (a duplicate, a misfiled note, a draft that went nowhere), delete it. The brain is in git. Anything you delete is recoverable from history. Do not hoard.

Rule of thumb: if you are not sure, archive. You can always pull a note back from archive/ to wiki/.

Invite the AI to help you grow it

You are not supposed to do all the maintenance yourself. The whole point of the brain is that it has an AI co-pilot. Every week or so, open the chat and ask one of:

  • "Look at the brain. What is missing that should be there?"
  • "What notes look like they should be split?"
  • "Which people in the brain should have richer pages?"
  • "Where are the dead ends in the graph?"

Treat the AI as a junior who maintains your filing system. You decide what matters, it does the work.