Zettelkasten in DEVONthink: Link Notes to Research Work
Build a small network of ideas in DEVONthink, connect each claim to its source, and carry reviewed conclusions into active research projects.
Updated 5 min read
You can build a Zettelkasten in DEVONthink by writing one useful idea per note, linking it to its source, and connecting it to other ideas with a sentence explaining the relationship. Start with a small group of notes and an index you can actually use while writing. The goal is reusable understanding, rather than a database full of saved documents.
Keep the research network in DEVONthink. When an insight changes active work, fenbs can hold the project decision and next step alongside a reference to the original note. That keeps your evidence separate from the instructions an assistant needs to continue the project.
How do you set up the first linked notes?
Choose one question you already need to answer, such as whether a design team should test a paper prototype before building a screen. Avoid importing your entire archive into a new structure. A handful of sources is enough to discover whether the workflow helps you think.
- Create a research database, or use an existing database you already maintain. Add three groups named Sources, Ideas, and Indexes. These are suggested organizational choices, not a required DEVONthink configuration.
- Save one source in Sources and record its author, title, original URL, and any page or section you will cite. Keep your reading notes with that source.
- Create a short editable note in Ideas. Give it a claim as its title, such as “Early prototypes expose unclear instructions.” Explain the claim in your own words.
- Copy the source item link and add it to the idea note. Then add a link to one related idea, with a sentence saying whether it supports, limits, or challenges the claim.
- Create an index note for your research question. Link to the best starting ideas and explain a sensible reading order. Return to this index when you begin the next writing session.
DEVONtechnologies describes editable annotations, item links, and WikiLinks on its knowledge-base feature page (opens in a new tab). Check the current handbook for your edition when choosing commands. The structure above depends on ordinary notes and deliberate links; you do not need to automate classification first.
What belongs in a source note versus an idea note?
A source note records what a particular document says. An idea note states something you understand and might reuse in a different project. Keeping them separate lets you change your interpretation without rewriting the evidence you originally consulted.
| Record | Put this in it | Example |
|---|---|---|
| Source note | Bibliographic details, precise locations, quotations clearly marked, and your reading questions. | A report describes participants missing a button on a prototype. |
| Idea note | One claim, your explanation, its scope, source references, and links to related claims. | A successful task test needs instructions that do not reveal the answer. |
| Index note | An entry point, selected links, and the question those notes help answer. | How should we prepare the first usability session? |
Use an explicit boundary sentence: “This source concerns a moderated test, so it does not establish what happens in an unmoderated study.” That makes the note more reusable than a broad claim with no limits. Add a question when the evidence is incomplete instead of smoothing the gap into certainty.
Should you use item links or WikiLinks?
Use a copied item link when you want a direct reference to a particular stored item. DEVONtechnologies explains that the address uses an item’s unique identifier and begins with x-devonthink-item://. Its item-link explanation (opens in a new tab) also describes links to specific locations in supported documents; that article flags older interface instructions, so use your installed help for the current controls.
WikiLinks are useful when you prefer to connect named ideas while writing. Whichever link style you choose, add readable text around it. “Contrasts with the assumption that testing requires finished screens” tells a future reader why the destination matters. A bare list of titles makes that reader reconstruct your reasoning.
Do not treat an item link as a public web page. Test it on the device where you intend to use it, and give collaborators a source citation they can access. A link stored in a shared tool does not give everyone the corresponding DEVONthink database or make its contents available to an AI assistant.
How do you find the right note when writing?
Build indexes around questions, rather than reproducing a textbook table of contents. “What makes feedback actionable?” is a better starting point for a design memo than a folder named Research. Include only the links that help answer that question, and write a short description beside each.
Try a retrieval check after each reading session. Close the source, open your index, and find the note that would support one paragraph in your current draft. If you cannot explain the connection, improve the note title or its relationship sentence before adding another source. Our second-brain collection covers the broader habit of keeping knowledge useful.
For AI-assisted writing, select the few conclusions the task actually needs. The distinction in what an AI second brain is helps: saving research and supplying useful working context are related jobs, but the complete archive is rarely the best instruction. Carry that selection into fenbs project context with the source references and the question the assistant needs to answer.
Connect the finding to the work it changes
Keep the full research note in DEVONthink and add a readable reference to Links and listings in fenbs. Record the reviewed conclusion in the project context, with its limits and the next research task. A connected assistant can read that work record without you pretending that a private database link supplies the underlying evidence.
Create a free accountFree for one person · No credit card · Works with Claude, ChatGPT, Cursor and more
What should you review before reusing a conclusion?
Reopen the source and check that your note still represents it fairly. Separate what the author observed from what you inferred. If a project uses a finding outside its original setting, record that as an assumption that needs checking.
- Can you locate the supporting passage without rereading the whole document?
- Does the note distinguish a quotation from your own explanation?
- Would the conclusion change under a different audience, task, or setting?
- Can the next person access the source, or do they need an approved extract?
When a person chooses what the project will do with the finding, record that choice separately from the note. That is where fenbs keeps project context and Decisions and rules. An evolving research idea should not silently become a standing instruction.
How to do this with fenbs
- Create a free account, then add the research project and its immediate question.
- Add the selected source references to Links and listings. Include a descriptive title and a citation or location that remains understandable when a private item link cannot open.
- Write a short context note explaining the reviewed finding, its limits, and the source that supports it. Record any choice a person made in Decisions and rules.
- Connect a supported assistant through Integrations. Ask it to read the project context before drafting and to save progress with the unanswered questions before it stops.
fenbs is free for one person, with no credit card. It does not import your DEVONthink database or replace its document library. If all you need is a private reading archive and linked notes, staying in DEVONthink may be sufficient; fenbs helps when the research feeds ongoing work across AI sessions.
Before switching assistants, save which notes informed the draft and what still needs verification. That gives portable AI memory a concrete meaning: the next session inherits a precise work record, while the original research keeps its own home.