Your note templates are an ontology (and why that helps)

A blank note is a bad place to start. Not because you don't know what to write — because you don't know what counts: is this a summary or a claim? Does the methodology go here or under limitations? Every one of those small decisions is friction, and multiplied across two hundred papers, it's the reason a pile of notes stops looking like a review and starts looking like a pile.

A template fixes this by answering the question before you start writing. And once you see what a template actually is — a fixed, named set of parts that every note of that kind will have — it's worth naming what that is: an ontology. Not in some heavyweight academic sense, just the plain one: an agreed vocabulary for what the parts of a thing are, so "the methodology section" means the same thing in every literature note you write, the same way a citation style defines what fields a reference needs before you ever fill one in.

The two templates, and what each one is for

ScholarThread ships two built-in templates:

Literature notes — for a single paper — have five sections: Summary, Core contribution, Methodology, Assumptions, Limitations. Each one prompts you with a one-line question when it's empty, so the template does double duty as a checklist: if you can't fill in "Limitations," that's a sign you haven't finished thinking about the paper, not just a sign the section is optional.

Concept notes — for an idea that spans multiple papers — have four: Definition, Why it matters, Open questions, Related concepts. This is a different shape on purpose. A literature note is about one paper; a concept note is about a thread that runs through several, so it doesn't have a "methodology" section — it has room for what's still unresolved and what else connects to it.

Both templates are editable, and both can be reset back to their shipped default any time you want a clean slate — the ontology is a starting point, not a cage.

Claims live on concept notes, not in a template section

Concept notes carry one more thing templates don't cover: claims. A claim is a distinct, ordered statement attached to a concept — "retrieval doesn't eliminate hallucination, it relocates it" — each one with its own evidence links back to the specific papers and passages that support it. Claims aren't a fifth section of the concept template; they're a structured list that sits alongside it, because a claim needs its own evidence trail, not just a paragraph of prose referencing sources in passing.

That distinction matters in practice: the template gives a concept note its fixed shape (definition, stakes, open questions, related ideas), while claims are where the actual argument — the specific, checkable statements the concept rests on — gets built up incrementally as you read more papers that bear on it.

Where AI drafting fits in

Two assisted-drafting tools sit on top of this structure, and both work the same way: they draft, you decide.

Populate a section from your papers. Next to any heading in a note, pick one or more papers from your library and ScholarThread drafts prose for that section — grounded in the specific passages it cites from those papers, not a general gloss on the topic. You review it, edit what needs editing, and explicitly accept it before it becomes part of your note. Nothing writes itself into a note without that step. This is a Pro feature.

Draft candidate claims for a concept. For a concept note specifically, ScholarThread can propose a set of candidate claims drawn from your library, each with a short rationale for why it's relevant. You look at each one and accept it individually into a real claim — with its evidence attached — or discard it. The model proposes; nothing becomes a claim in your concept until you say it does.

Both tools exist because the hard part of note-taking was never typing — it's staring at an empty section deciding what belongs there. Grounded drafting gives you a first pass to react to instead of a blank prompt to fill.

Doing this in ScholarThread

Start a new literature note from a paper, or a concept note from an idea, and the template's sections are already there, with their prompts, waiting to be filled in — or handed to "Populate with AI" if you want a grounded first draft to edit instead. For a concept you're building up over several papers, add claims as you find evidence for them, or ask ScholarThread to draft candidates and accept the ones that hold up.

The template isn't there to constrain what you think. It's there so that six weeks and two hundred papers later, every note still means the same thing it meant on day one — which is exactly what a repeatable literature-review workflow needs underneath it.