How to Run a Literature Review in Heptabase Whiteboards
Heptabase is a visual note-taking tool built around an infinite whiteboard. Instead of stacking notes in folders, you write atomic cards and arrange them in space. For research work, that turns a mountain of PDFs into a map you can read at a glance.
The tool ships with a split-screen PDF reader and a context-aware AI chat, so the workflow runs from reading to synthesis without leaving the app. Below is a setup that survives a long project.
What makes Heptabase different from Notion or Obsidian?
Spatial arrangement. Notion is a database with pages, Obsidian is a graph of linked files. Heptabase is a canvas where position carries meaning.
That sounds cosmetic until you have 200 notes. In a folder system, you sort by tag and hope. On a whiteboard, you physically group “methods” in one corner, “open questions” in another, and the structure is visible without opening anything.
How do you start a research project?
One whiteboard per project, named with a date so you can archive it later.
- Create the whiteboard, then add three sections with the Section tool: Sources, Claims, Questions.
- Drop every PDF into the Sources section. Heptabase imports them into the built-in reader.
- Highlight inside the reader, and each highlight becomes a card you can drag onto the canvas.
- File each highlight card under Claims or Questions, depending on what it does.
Do not try to be tidy in the first week. The point of the board is to dump, then arrange.
How do you handle PDFs and highlights?
The split-screen reader is the feature that makes the workflow viable. You read on one side, and pulled quotes land on the other as cards.
Two habits help:
- Highlight sparingly. Ten highlights per paper, maximum. If everything is highlighted, nothing is.
- Rewrite in your own words on the card. Heptabase keeps the source link, so you do not lose the quote, but your note should be your sentence. Future you reads the note, not the paper.
How should you use the AI chat?
Attach the whiteboard or specific cards to the chat, then ask it to summarise or find gaps. The key is context selection.
Good prompts:
- “These twelve cards describe competing methods. Group them into three families and name the axis they differ on.”
- “Which of these claims contradict each other, and on what specific number?”
Bad prompts: “summarise my research.” Attaching the whole board produces a generic summary that reads well and says nothing. Attach 10 to 20 cards, not 200.
The AI only sees what you attach, and it grounds answers in your cards rather than inventing citations. That is the difference between this and asking a chatbot about a topic it half-remembers.
What does it cost and where does it hit limits?
The paid plan is around $8.99 per month billed annually, and the free tier is limited enough that serious projects outgrow it within a month.
Limits worth knowing:
- Whiteboards get slow past roughly 300 to 400 visible cards. Split into sub-boards once you cross that.
- No native mobile editing of whiteboards; the mobile app is mostly a reader.
- Collaboration is limited. It is a single-user tool with light sharing, not a multiplayer doc.
Who is this workflow wrong for?
Teams. If three people need to edit the same research, use Notion, Google Docs, or a proper reference manager.
People who think linearly. If your writing process is “outline, then fill in”, the canvas adds friction. Heptabase rewards people who think in fragments and connect later.
Quick Summary
- Heptabase is a whiteboard-first note tool: atomic cards arranged in space, plus a split-screen PDF reader.
- Start with three sections (Sources, Claims, Questions) and dump first, arrange later.
- Highlight ten quotes per paper at most, and rewrite each on the card in your own words.
- Attach 10 to 20 cards to the AI chat, never the whole board, or summaries go generic.
- Around $8.99 per month annually; boards slow past 300 to 400 visible cards; not built for team editing.
