Research you can see,
beside the chapter it feeds.

Organize by the shape of the book rather than by the type of the material. An interviews folder, a photos folder, and a papers folder feels tidy and is almost useless while writing, because at the moment you need something you are thinking about a chapter, not a file type. Cluster by chapter, theme, or argument instead, and let a source that serves three chapters sit visibly between them. Keep three things separate and always visible: the material itself, your own synthesis notes about what it means, and the open questions you still have to resolve. Record the citation at the moment you find a source. And keep a chronology if the book has any temporal spine at all, because time is the index you will reach for most.

Record the source at the moment you save the material, not later. Put the citation detail directly on the card that holds the material: author, title, date, page or timestamp, and a link where one exists. This sounds obvious and is the single most common failure in book research, because finding a source again eighteen months after you read it can take an afternoon per fact. If you are writing something with formal citation requirements, a dedicated reference manager like Zotero is worth running alongside, with its key on the canvas card. Storyflow's link cards keep online sources live so you can re-check anything before publication.

Not during early access. Storyflow is paid-only right now, with plans starting at $9.99 a month, and a Free plan for solo use launches before the end of 2026. Anyone a paid member invites to a board joins free today, so a co-author, researcher, or editor can join without paying. If you want a free option this minute, Obsidian is free for personal use and genuinely excellent for linked research notes, and Zotero is free and remains the best free citation manager available.

The AI keeps your whole canvas in context, which is what makes it useful here rather than as a chat window. You can ask what your sources say about a particular claim, where two accounts disagree, which chapter has the thinnest supporting material, or what themes recur across a set of interviews. Because it works from the board you built rather than from general knowledge, the answers are about your book. It is a research assistant that has read your files, not a search engine, and like any assistant its output is worth checking against the source before you publish a claim on it.

Yes, and the pattern is the same even though the material differs. Novelists doing period, technical, or location research need it beside the scenes that depend on it, and the failure mode is identical: you researched how a particular job worked in 1890 and cannot find the note while writing the scene. Keep the research clusters next to the plot cards they support on the same canvas. Worldbuilding, character research, and location reference all work the same way, and the moodboard for the novel's atmosphere can sit on the same board.

Make the gaps visible rather than the volume. A board where every chapter has a research cluster next to it shows you immediately which chapters are thin and which are over-served, and over-researched chapters are usually the ones you were avoiding writing. Keep the open questions cluster small and specific, and treat anything not attached to a chapter or an argument as material you have not yet justified collecting. Seeing the whole board at once tends to make the imbalance obvious in a way that a growing folder never does.

Yes. Invite them to the board and they can add material directly to the canvas, placed near the chapter or theme it belongs to. Anyone a paid member invites joins free today, so a researcher does not need their own plan. This is a meaningful improvement over receiving research as documents, because an assistant who can see the book's structure files things where you will actually look for them, and you can see what has arrived without opening anything.

You can grab frames from YouTube and Vimeo directly onto the canvas, which covers documentary footage, recorded lectures, and archive film. Note the timestamp beside the frame so you can return to the exact moment. For interview audio, the common pattern is to transcribe elsewhere and drop the transcript onto the board as a document, with the audio file linked from a card beside it. Storyflow is not a transcription tool, so the transcription step happens in whatever you already use.

That is the real test, and it is mostly about whether you can reconstruct your own thinking. Two things matter more than the tool: keep synthesis notes attached to the material they came from, so returning to a source in year three tells you why it mattered as well as what it said, and record citations at the point of capture. Boards are unlimited with no object cap, so a multi-year project does not force you to prune. For anything you must be able to open in decades, keep an archival copy of primary material outside any single tool, which is prudent regardless of what you are using.