Roam Research Zettelkasten: Make Your Note Links Useful

Build a linked-note system around ideas you can explain and sources you can find again. Turn the useful connections into research you can actually use.

Updated 5 min read

A Roam Research Zettelkasten is a collection of short idea notes connected by meaningful links, with references that let you check where each idea came from. Start with one research question, write each useful idea in your own words, and explain why it connects to an existing note.

The goal is to recover a useful line of thought when you need it. A dense network of titles does not establish an argument. When research becomes project work, fenbs can retain the selected references, agreed conclusions, and next steps that different AI apps need to share.

How do you build the first connected notes?

  1. Choose a live question, such as whether a shorter intake form would help customers finish an application. Give the question a home note.
  2. Read one relevant source. Save its title, author, link, and the page or section that supports the point you want to use.
  3. Create an idea note with a title that states the idea. Explain the claim, its limits, and the evidence in a few sentences.
  4. Link it to the question and one related idea. Add a sentence describing the relationship: supports, challenges, explains, or provides an example.
  5. Return to the question note and write what the new connection changes. Leave unanswered questions visible.

Roam organizes pages into blocks, the individual pieces of an outline. Its official syntax reference (opens in a new tab) documents page links such as [[Intake form research]] and references to individual blocks. Use page links for an idea you want to revisit as a whole; use a block reference when an exact existing passage matters.

Begin with a small working set. You do not need to organize your entire reading history before making the first useful connection. For source-specific capture, the separate guides cover book notes and daily notes.

What should each idea note contain?

Give an idea enough explanation to make sense outside the moment you wrote it. A title such as “Fewer questions can reduce uncertainty only when instructions remain clear” is more useful than “Forms.” It tells you what the note argues before you open it.

  • Claim: the point you think is worth keeping, expressed in your own words.
  • Evidence: the source and precise location that support it, separate from your interpretation.
  • Boundary: when the idea may not apply or what the evidence does not establish.
  • Connection: another note and one sentence explaining the relationship.
  • Use: the question, draft, or project where this idea could matter.

For example, your hypothetical form note might distinguish removing an unnecessary question from removing an explanation customers need. Those are different changes with different possible effects. Keeping that distinction in the note prevents a later summary from turning “reduce unnecessary questions” into “make every form shorter.”

A quotation is evidence to interpret, not a substitute for your own note. Keep quotation marks and a location whenever you preserve exact words. Cornell University Library’s research guidance (opens in a new tab) emphasizes evaluating sources and citing them so another reader can find them.

Which links are worth adding?

Add a connection when it helps answer a question or follow an argument. Sharing a broad topic is a starting clue, but it is not always enough. Two notes about customers may concern different populations, different decisions, or conflicting evidence.

RelationshipUseful connecting sentence
SupportsThis finding supports the intake note because it examines the same point of confusion.
ChallengesThis example challenges the shorter-form idea because customers still need an eligibility explanation.
Depends onThis recommendation applies only if customers can get help before submitting.
Suggests a questionTogether these notes suggest testing the instructions separately from the number of fields.

Avoid creating several names for the same idea. Search your existing titles before creating another page, and prefer a name you would naturally use when looking for it. If you cannot explain a connection, leave it as a question rather than writing a confident relationship you have not established.

How do you review the graph without chasing its shape?

Review your graph, meaning the connected collection of notes, from one working question outward. Open the notes linked to that question and read the relationship sentences. Can you reconstruct the reasoning without remembering the original reading session?

Look for three problems: an isolated claim you cannot trace to evidence, a popular concept that connects everything without explaining anything, and an apparent agreement that actually joins conflicting claims. Repair the text before adding more links. A smaller collection of understandable relationships is easier to use than a larger collection of unexplained ones.

End the review with a short synthesis: what you currently believe, what supports it, and what would change your mind. Write the synthesis as your present understanding, not as a permanent truth. Keep source notes intact so a future correction has somewhere to start.

What should an AI assistant receive from your research?

Give an assistant the question, selected idea notes, source locations, and the distinctions it must preserve. Ask it to identify unsupported claims before drafting an answer. A note link is a pointer; it does not prove the assistant has opened or understood the destination.

Roam also provides an official AI connection (opens in a new tab). Its documentation describes a desktop-based connection and recommends backing up the graph and reviewing planned changes. Follow the current instructions if you use it; a writing assistant should not reorganize your research merely because it can edit notes.

Keep tentative research separate from an approved project decision. “These sources suggest a simpler form” and “the owner approved removing the company-size field” have different authority. The shared AI brain guide explains that shared work record, while portable AI memory covers keeping useful context across apps.

How fenbs helps

Keep the references behind the agreed answer

Add the selected research links to Links and listings in fenbs, and explain their relevance in project context notes. Record a person’s approved conclusion in Decisions and rules. Your connected assistants can then start with the project answer and its references while the exploratory notes stay in Roam.

Create a free account

Free for one person · No credit card · Works with Claude, ChatGPT, Cursor and more

How to do this with fenbs

  1. Create a free account and add the project your research supports.
  2. Add selected source and note links to Links and listings. Include enough explanation in an AI context note to identify what each source supports.
  3. Record what a person actually decided in Decisions and rules. Keep unresolved research questions open and add tasks for the work they require.
  4. Connect a supported AI app from Integrations. Ask it to read the project context before using the selected research, then save Where we left off before stopping.

fenbs does not visualize a personal knowledge graph or import your Roam collection. Keep using Roam if the main job is exploring relationships between personal notes. Use fenbs when a selected conclusion needs a shared owner, task, or handoff; its personal account is free for one person, no credit card.

For your next review, choose one research question and follow its links to an actual source. Repair anything you cannot explain, then write the next action. The second-brain library has related approaches when your main difficulty is capturing or finding notes rather than connecting ideas.

Questions people ask.

Do I need a numbering system for every Roam note?

You can use stable identifiers if they help your work, but begin with clear titles and meaningful links. An elaborate numbering scheme will not repair a note whose claim or source is unclear.

Can one idea note link to several sources?

Yes. Explain whether the sources agree, disagree, or support different parts of the idea. Keep their locations separate so you can check each claim.

Should I delete notes that no longer seem correct?

Often it is more useful to mark the problem and link to the correction. Old reasoning may explain why you once chose an approach, provided readers can see that it has been superseded.

Will an AI find every useful connection for me?

An assistant can suggest relationships among the material it can access, but you still need to check their meaning and evidence. Treat suggested links as candidates rather than established conclusions.

Start with one thing.

Free for one person · No credit card · Works with Claude, ChatGPT, Cursor and more

There is nothing to set up first. Write one line and you’ve started.