Obsidian Zettelkasten: Link Notes for AI-Assisted Research

An Obsidian Zettelkasten is a collection of focused notes connected by links that explain how ideas relate. For AI-assisted research, give every evidence-based claim a route back to its original source, then check any generated synthesis against those sources. fenbs can keep the research question, approved scope, and follow-up tasks separate from the note archive.

Start with one question and a few sources you can read closely. The useful result is an argument you can inspect, including its uncertainty. A large graph of links does not establish that a claim is true or that an assistant interpreted the evidence correctly.

How do you build the first linked research notes?

Use a small loop that separates collecting evidence from developing an idea. Keep rough reading notes if they help, but create a reusable idea note only when you can explain what it says. The link between notes should carry a reason, not just a shared subject.

1. Write one research question and the limits of the investigation, including the population or situation you mean.
2. Create a source note for each document you actually read. Record its title, author or organization, location, and relevant passage or page.
3. Write a separate idea note for one claim in your own words. Link it to the source note and state what the evidence does and does not support.
4. Link the idea to another note with a sentence explaining the relationship, such as contradiction, a condition, or a useful example.
5. Return to the research question and list the checks or missing evidence needed before drawing a conclusion.

For example, say you are examining why people leave an online application unfinished. One idea note might say that unclear document requirements can interrupt completion. Keep that claim attached to the specific research behind it, rather than presenting it as the explanation for every abandoned application.

What belongs in a source note versus an idea note?

A source note identifies and describes evidence from a particular work. An idea note develops your understanding of a claim, perhaps using several sources. Keeping these jobs distinct lets you revise your interpretation while preserving what the source actually said.

Obsidian properties↗ (opens in a new tab) can hold text, dates, links, and other structured information about a note. You might use properties for a note’s type, source identifier, or review date. Those are fields you maintain; adding them does not validate the underlying evidence.

Title: Unclear document requirements can interrupt an application
Claim: A person may stop when they cannot identify the required evidence.
Basis: Link to the source note and relevant section.
Scope: Describe the participants and application studied.
Interpretation: Clearer instructions may help; that intervention is untested here.
Connection: Contrast with the note about technical submission failures.
Next check: Find evidence that separates confusion from missing documents.

How should you link notes so the connections mean something?

Obsidian supports links to notes, headings, and blocks in its internal links documentation↗ (opens in a new tab). Use the most precise location that is helpful. A link to the exact discussion of a limitation saves the next reader from searching a long source note.

Write a sentence around each important link. “This challenges the explanation based on technical failures because participants described uncertainty before submission” carries reasoning. A list of related filenames only says that you noticed some association.

The Backlinks core plugin↗ (opens in a new tab) shows notes that link to the active note. Use that view when revising a claim to find arguments that may depend on it. A backlink reveals a connection you recorded; it does not confirm the connected argument is sound.

Avoid linking every note containing the same word. Prefer relationships that help answer a question or expose a disagreement. A small network with explanations is easier to review than a dense collection whose links have no stated meaning.

How can an AI help you synthesize the notes safely?

Choose a small set of notes and the relevant source passages you are permitted to share. Ask the assistant to identify claims, supporting evidence, contradictions, and gaps before drafting a conclusion. Require it to distinguish the source’s statement from your interpretation.

Use only the notes and passages supplied.
For each proposed conclusion, identify its supporting source passage.
State the population or conditions the evidence concerns.
Separate direct support, inference, contradiction, and missing evidence.
Do not invent references or treat repeated summaries as independent sources.
Return open questions before drafting the final argument.

Read the cited passages yourself. Check whether a source supports the whole claim, a narrower claim, or only a related observation. Two notes repeating the same report are not independent evidence. If the assistant cannot identify a passage, keep the claim unresolved.

The research workflow guide covers the broader research process; the AI search guide covers finding material. This linked-note workflow begins once you have sources to inspect, and it cannot replace that inspection.

Where should the research question and next work live?

Keep a clear current record of the question, intended output, human decisions, and remaining work. The note archive may support several investigations, so it should not be the only place that says which one you are pursuing now. A new assistant needs that distinction before it starts collecting more material.

The AI brain guide explains this continuing work record, and the AI at work guide covers choosing a bounded task. For this project, a bounded next task might be checking one unsupported explanation rather than “finish the research.”

When reviewing Google AI tools for research, check how you will supply material you are allowed to share. A note stored locally is not automatically available elsewhere. Record what was actually reviewed, especially when an assistant only saw excerpts.

fenbs keeps source references alongside project context, giving the next assistant a current question and follow-up tasks while the detailed note network stays in Obsidian.

How to do this with fenbs

1. Create a free account, then add a project with the research question and intended result. fenbs is free for one person, no credit card.
2. Keep source and idea notes in Obsidian. Add relevant source references in Links and listings and explain their significance in AI context notes.
3. Add verification tasks and record the scope a person approved in Decisions and rules. Keep unresolved choices open for a person.
4. Connect a supported assistant, ask it to read project context before work, and save checked results, uncertainty, and next steps before stopping.

fenbs does not synchronize your Obsidian vault or automatically import its notes. It is not the right fit if all you need is private linked-note writing. Use it for the research work around the archive, and see the second brain hub for related organization approaches.