The literature lives in Zotero. The fieldnotes live in a notebook. The hypotheses live in a Google Doc. By the time I sit down to write, I've lost the thread between them — and the thread was the work.
The reading is fine. The fieldwork is fine. The drafting is fine. What breaks is the seam between them. We didn't build another reference manager — we built the part of the workflow that was missing between the citation, the highlight, and the sentence you eventually have to write. Some people will call this a research notes app. We call it the missing function of academic memory.
A research project is not three jobs that meet at the end. It is one continuous act of synthesis — the literature pulling on the fieldwork, the fieldwork talking back to the hypothesis, the hypothesis re-shaping which papers you read next. When you split this act across three tools, you don't lose data. You lose the thread.
The citation manager holds the bibliography but not what you thought when you read it. The notebook holds the fieldnote but not the paper that made the question worth asking. The Google Doc holds the draft but not the precedent that should already be cited. So every time you open the project, you do the synthesis again — from scratch, in your head, while three tabs blink at you.
That isn't a tool problem. It's a seam problem. And the seam is where the intellectual work actually happens.
Knovya holds the literature, the fieldnotes, the hypotheses, and the drafts in one archive. Citations link directly to the passages that informed them. Atomic notes surface the neighbors they belong with — paper to paper, paper to fieldnote, fieldnote to claim — without you maintaining the index.
Hybrid search reads keyword and meaning at once, so a half-remembered phrase from chapter three of a paper you read in February finds its way back to you in November. AI Memory surfaces the prior reading you almost forgot to cite. When you sit down to draft, the synthesis is already on the page, waiting to be argued.
We don't call ourselves a research notes app. We call it the missing function of how academics already work.
The argument holds the page before you write the first sentence.
Three moments from a typical research week. Pick one — the archive lights up the part of itself it would actually use. No live data, no signup; the moves are real, the notes are illustrative.
A paper imports from Zotero — citation, abstract, and annotated PDF arrive as one note.
It finds the three prior papers it shares vocabulary with — and the fieldnote that quoted one of them.
AI extracts the paper's claims as atomic notes in your own paraphrasing — each with the page reference attached.
The next time you draft, the paper is already in the citation queue — BibTeX entry waiting beside the paragraph.
Ahrens called these the four moves of academic note-taking. We made the archive carry the four of them so the writing carries the argument.
Knovya doesn't ask you to follow a method — the archive does. Here's which features carry which move, mapped to the elements you'll find on the periodic table at /features.
Bring the paper, the interview, the conversation in — with the source intact.
arXiv, JSTOR, any URL — clipped as a note with the citation attached.
A Claude or ChatGPT thread, saved as a real entry your archive will keep.
Fieldnote dictation — interview, lab, archive — arrives structured.
The archive wires itself — backlinks, semantic neighbors, the citation graph that grows underneath.
Every reference visible from both sides — paper to claim, claim to chapter.
A view of how your literature, fieldwork, and argument connect.
Which papers are doing the work — ranked by how many ideas pull on them.
The right precedent finds you. Hybrid search across reading, AI memory across projects.
Outline a lit review, summarize an interview, extract themes across sources.
Keyword and meaning together — finds the half-remembered passage.
Past papers and prior projects surface alongside the new one.
Forgotten reading returns when its topic comes up in a new draft.
The archive is the start of the chapter — drafts, citations, exposed to every AI you write with.
Claude, ChatGPT, Cursor — they read your literature natively while you draft.
Publish a preprint thread or chapter, keeping the rest of the archive private.
The phrase research notes is recent. The work is old — every scholar who has ever held more reading than memory could carry. The shape of the answer has changed; the question hasn't. How do we keep what we read close to what we're writing?
Before the typewriter, before the photocopier, scholars wrote one idea per slip and shuffled them into arguments. Walter Benjamin's unfinished Arcades Project — tens of thousands of citations, paired with marginalia, never bound into a book — is the visible monument of an invisible practice every researcher of the period kept.
The first reference manager that mattered. EndNote put the bibliography into a database and let it cite itself into a paper. Synthesis stayed in the head; what got automated was the formatting of the synthesis you'd already done.
The Center for History and New Media released Zotero as open source — a Firefox extension that captured a paper from any catalog page in one click. Citation management stopped being a product and became the air researchers breathe.
Mendeley made the PDF the unit. Sync across devices, social discovery, group libraries. Elsevier acquired it in 2013; the desktop app was sunset in 2022. The lesson the field took away wasn't about features — it was about stewardship.
Ahrens translated Niklas Luhmann's slip-box for the digital researcher: literature notes paraphrased in your own words, atomic permanent notes linked to neighbors, project notes that gather toward a draft. The book quietly seeded a decade of academic adoption of Roam, Obsidian, and Logseq.
NotebookLM (out of Project Tailwind, 2023; mainstream in 2024) let you upload sources and ask the corpus questions in natural language. Every answer cited the passage that produced it. The shape was new; the limit was that the corpus is per-notebook — the reading doesn't follow you to the next project.
We built the part the index card pointed at, the citation manager indexed, Ahrens taught, and NotebookLM almost reached: one durable archive that holds the literature, the fieldnotes, the hypotheses, and the drafts at once. NoteRank ranks the papers actually doing the work; MCP exposes the archive to Claude, ChatGPT, and Cursor while you write. The seam, finally external.
Every tool in this category wagers on one stage of the research cycle — usually one, sometimes two. The honest comparison isn't features. It's which move each app decided to be best at, and which it's leaving to you.
The bet Citations done right. Open source, open standards, group libraries, browser capture. The bibliography is a public good, kept by the people who use it.
What's left to you Synthesis. Zotero indexes what you've read. The connections between the papers, the fieldnotes that reframe them, the argument that cites them — those live somewhere else.
The bet The PDF as the unit. Annotate, sync, share. Originally a startup with social discovery; now part of Elsevier's research-information stack.
What's left to you Trust about the long term. The desktop app was sunset in 2022. The PDF is a fine unit; the question is whose roadmap holds it.
The bet Ownership and durability. Plain markdown on your machine, plugin-extensible, outlives any company — the digital slip-box for the Smart Notes generation.
What's left to you The linking discipline. The vault is yours to maintain. Plugins help — Citations, Dataview, Zotero integration — but the connective tissue is your daily labor.
The bet A reading partner. Upload sources, ask them questions, get cited answers. Audio Overviews and Deep Research extend the surface; the corpus stays the focus.
What's left to you Archive permanence. Each notebook is its own corpus. The reading you do today doesn't follow you to the project next semester; the synthesis is per-session.
The bet The seam is the product. Read, link, synthesize, and argue belong in one durable archive — and that archive should expose itself to Claude, ChatGPT, and Cursor natively while you write.
What's left to you The argument. Citation, linking, surfacing, and prior-art recall are on the system. The thesis you're trying to defend — that part stays yours.
We didn't pick a stage to be best at. We picked the seam between them.
Research notes only earn their keep if they're reachable in the seven seconds between finishing a paragraph and forgetting why it mattered. Knovya works on every surface where those seconds exist.
In the hallway outside the room, in the car back to campus — dictate the fieldnote; arrive at the desktop with a structured entry already waiting.
The graph view shows which sources, fieldnotes, and atomic notes are doing the most work for a given chapter — and which threads are starting to braid into a new argument.
The Chrome extension turns any arXiv, JSTOR, or publisher page into a note with the citation parsed, the BibTeX entry queued, and the semantic neighbors already wired in.
Every note carries its source. Select a tag, a chapter, or a project; export to BibTeX, APA, Chicago, or MLA — paste straight into LaTeX, Word, or your reference manager.
Research notes aren't a feature — they're the shape the whole archive takes when its parts cooperate around a question. Here's the constellation around this page.
Outline a lit review, summarize an interview, extract themes across sources.
See featureKeyword and meaning together — finds the half-remembered passage.
See featureA view of how literature, fieldwork, and argument actually connect.
See featureClaude, ChatGPT, Cursor — they read your literature natively while you draft.
See feature"I take great notes during the lecture. By the exam, I can't tell what's load-bearing."
Notes that hold what mattered — not what you wrote down, but what the course was actually about.
See the workflow Problem 02 · Chapter I"Everything I read is somewhere. I can't find any of it when I need it."
The right note, surfaced at the right moment — without you remembering you needed it.
See the workflowThe practice underneath research notes — capture, retrieval, the recall-to-recognition shift.
Read the primer DefinitionHow papers, fieldnotes, and arguments actually link — and why backlinks are the academic slip-box's digital descendant.
Read the primerResearch archives aren't built in one sitting. They're built one citation, one fieldnote, one atomic claim at a time. The argument starts compounding the moment the seam closes.
Or scroll back to the diagnosis.
Eight questions we keep getting. If yours isn't here, the contact page reaches us directly.
Research notes are the written record of how you read, interrogate, and connect sources. In academic practice, the term covers four overlapping artifacts: literature notes (your paraphrased reading of a paper), fieldnotes (qualitative observations from interviews, ethnography, lab work), atomic notes (one idea per note, in your own words, citing the source), and project notes (the working argument that draws from the other three). Knovya is built for the act that ties them together — synthesis.
The Smart Notes method (Sönke Ahrens, after Niklas Luhmann) is the most reliable rhythm: read with a pen, write a literature note in your own words with the citation attached, then promote the ideas you actually engaged with into atomic notes — one claim per note, linked to neighboring ideas already in your archive. Knovya automates the linking step (semantic neighbors surface as you type) so the discipline that wears researchers down is carried by the system.
Stop organizing by folder, start organizing by connection. The literature you've read, the fieldnotes you've taken, and the arguments you're forming all live in one archive; backlinks and a knowledge graph let you see which papers are pulling toward which chapter. The folder you would have used was a guess about what you'd need later; the graph reflects what you actually returned to.
Fieldnotes are the on-site, in-the-moment record a researcher keeps during fieldwork — interviews, ethnographic observation, lab sessions, archival visits. They differ from literature notes in two ways: they capture what was happening (descriptive) and what you noticed about it (reflective), and they're written under time pressure. Knovya supports voice-captured fieldnotes that arrive as structured entries on your desktop, ready to be linked to interviews, sources, and emerging themes.
It depends on what you split your work across now. Zotero is the gold standard for citation management; Obsidian for local-first markdown linking; NotebookLM for source-grounded Q&A. Each is excellent at one stage of the research cycle. Knovya's bet is different — it holds the literature, the fieldnotes, the hypotheses, and the drafts in one archive that exposes itself to Claude, ChatGPT, and Cursor through MCP. The synthesis step doesn't get handed to you in pieces.
Zotero is a citation manager — strong on bibliography, weak on synthesis; you still write the connections yourself. Obsidian is local-first markdown — strong on ownership, weak on automatic linking; the slip-box is yours to maintain daily. NotebookLM is closed-corpus Q&A — strong on reading inside one notebook, weak on archive permanence; the corpus is per-notebook and ephemeral. Knovya holds everything in one durable archive with bidirectional links, NoteRank, AI Memory, and citation export to BibTeX, APA, and Chicago.
Yes. Every Knovya note carries its source metadata; the citation export view turns selected notes into BibTeX, APA, MLA, or Chicago entries on demand, ready to paste into LaTeX, Word, or your reference manager. The atomic notes themselves export as markdown with frontmatter — citations, tags, backlinks, and the date you wrote them — so the archive is portable, not a lock-in.
Pro and Team plans include note-level end-to-end encryption (AES-256-GCM); encrypted notes are not searchable or embeddable on the server. The Free tier uses transport encryption. Login is hardened with 2FA, device fingerprinting, and anomaly detection. Pre-publication work, IRB-sensitive transcripts, and fieldnotes covered by participant agreements stay where you put them.