Heptabase

Heptabase is a visual knowledge management tool built specifically for students, researchers, and anyone who needs to synthesize large amounts of source material into connected understanding. Founded by Yu-An (Alan) Chan and backed by Y Combinator and Kleiner Perkins, the product launched in 2022 with a simple premise: your notes should live on a spatial canvas, not in a linear list. Every piece of information you save becomes a "card," and cards can appear on multiple infinite whiteboards at once, so a single research note can belong to three different projects without duplication. The company reached $1.2M ARR with a team of ten by late 2024, growing steadily while staying profitable on a lean headcount.

The core feature set covers the full research workflow: a block-based card editor with bi-directional linking, PDF import with native OCR and highlight-to-card conversion (added September 2025), YouTube and podcast transcript import, a web clipper browser extension, daily journal with due task display, and AI chat that queries your entire knowledge base by semantic search and cites specific cards and PDF page numbers. The December 2025 update added AI actions directly on whiteboard cards (translate, summarize, create mind map from a card) plus Model Context Protocol integration, so tools like Claude and ChatGPT can read and write to your Heptabase knowledge base. Three paid tiers (Pro, Premium, Premium+) differ primarily by AI credit allotment and model access. There is no free plan after the 7-day trial.

What Heptabase actually is in April 2026

Heptabase sits at the intersection of personal knowledge management (PKM) and visual thinking tools. The product's architecture is built around two synchronized views: a flat card library (every note you've ever created, searchable globally) and infinite spatial whiteboards (curated collections of cards arranged spatially to represent a project, argument, or topic). The whiteboard is not a bolt-on; it is the product's primary interface for working with knowledge.

As of April 2026, the feature surface includes: a block-based card editor supporting markdown shortcuts, images, code blocks, and embeds; bi-directional linking with backlinks visible in a right sidebar; PDF annotation that automatically creates linked cards from highlights (with full OCR parsing introduced in v1.74.0, September 2025); YouTube and podcast transcript import with timestamp-linked cards; a "Research a new topic" flow that creates a dedicated whiteboard, imports sources, and surfaces AI chat scoped to that whiteboard; a right-sidebar AI suggestions panel that surfaces related cards during writing; and an AI Tutor for structured, quiz-style learning sessions. The mobile apps (iOS and Android) support full card editing and a dedicated Task app; whiteboard manipulation on mobile is functional but limited compared to desktop.

On January 2024, Heptabase won Product Hunt's 2023 Golden Kitty Award in the Personal Productivity category, the platform's first major community validation. The same month brought the announcement via co-founder Wu Ping Ju that named Heptabase as one of the standout personal productivity tools of the year. The Golden Kitty drove a sustained spike in signups and positioned Heptabase as the PKM category's visual-first challenger to Obsidian.

"The level of detail and understanding of human-computer interaction that went into this product is insane." - Glenn Gregor, Product Hunt, 2023

Where Heptabase sits versus Obsidian Canvas and Miro

The two most common comparison points are Obsidian Canvas and Miro. They represent opposite failure modes Heptabase was designed to avoid: Obsidian Canvas is powerful but plugin-mediated, while Miro is polished but retains nothing you learn.

Heptabase vs. Obsidian Canvas: Obsidian's Canvas feature is a JSON file that references existing Markdown notes stored as local files in folders. Every advanced capability, including PDF annotation (via the Annotator plugin), Zotero integration, and custom database views, requires installing and maintaining community plugins. Heptabase has no plugins at all; the full feature surface is built-in, including native PDF OCR parsing, YouTube transcript import, Zotero and Readwise sync, web clipper, and AI chat. The tradeoff is control: Obsidian is free, local-first, and open to any plugin; Heptabase is $8.99/mo minimum, cloud-synced, and not extensible. For PDF-heavy academic workflows, Heptabase's built-in highlight-to-card conversion is concretely faster than Obsidian's plugin chain. For users who need local file storage, version control of notes, or plugin ecosystems, Obsidian wins.

Heptabase vs. Miro (and tldraw): Miro is a team collaboration whiteboard for facilitated sessions: sprint retrospectives, design reviews, journey mapping. Sticky notes on a Miro board do not persist as searchable knowledge objects; they exist for the duration of a workshop and are then archived. Heptabase's cards are persistent, searchable, AI-queryable notes that happen to appear on a canvas. Miro's AI (Miro Assist) generates sticky note suggestions and diagram templates. Heptabase AI cites specific card content and PDF page numbers. Miro is per-seat priced (starting $8/member/mo) and built for teams of 5+. Heptabase is per-user and built for the individual researcher. tldraw sits further away still: it is an MIT-licensed open-source drawing library for embedding collaborative canvas in custom applications, with no knowledge management layer, no notes, no search, and no persistence model beyond raw canvas state.

"Love the product! It's been game changing when brainstorming to be able to put my thinking in a mind map." - Guillaume Luccisano, Product Hunt, 2023

How AI actually works inside Heptabase

Heptabase's AI is not a generic chatbot wrapper. It is scoped to your knowledge base. When you open AI chat, it retrieves contextually relevant cards using both keyword and semantic search across your entire space, reads those cards, and answers with citations linking back to specific cards, PDF pages, or video timestamps. The difference from general AI chat is that answers draw only from material you have saved, reducing hallucination on your domain and surfacing connections you may have forgotten you made.

Three AI access modes exist: global AI chat (queries your whole space), whiteboard-scoped AI chat (queries only the cards on a specific whiteboard, useful for focused project work), and card-level AI actions (hover over a whiteboard card to get a menu of AI operations: translate, summarize, create a mind map, run a custom saved prompt). The December 2025 MCP launch extended this model outward: Claude, ChatGPT, and other MCP-compatible AI services can now read from and write to your Heptabase space directly. You can ask Claude to summarize a discussion thread and save the result as a card in today's journal, without switching applications.

Plan gating applies to AI. The Pro plan (the cheapest paid tier) provides 100 AI credits per month and limits AI chat to Gemini series models only. The Premium plan provides 1,800 credits per month, adds OpenAI and Anthropic model access, and includes unlimited basic AI chat (Gemini Flash, GPT mini) once credits expire. Premium+ provides 8,100 credits per month. Users with intensive AI workflows report that the Pro plan's 100-credit ceiling runs out within a few working days of heavy PDF processing and AI chat use.

The friction users keep raising

Performance on large whiteboards is the most consistent complaint in community reviews. Once a whiteboard accumulates 100 or more cards, particularly cards with embedded PDFs or many highlights, users report noticeable lag. One reviewer noted plainly: "Whiteboards with 100s of cards and pdf highlights are slow and laggy." The team's fast feature release cadence appears to have come at some cost to performance optimization.

Buggy feature rollouts are a secondary friction point. New capabilities ship weekly, but community reviews note that initial releases often have rough edges. As one Product Hunt reviewer wrote: "Features are consistently buggy. Application performance doesn't seem to be a priority." This is the tradeoff of a 10-person team shipping aggressively. The MCP integration, for example, initially required manual configuration steps that were confusing for non-technical users.

The mobile experience gap is real. The iOS and Android apps support card editing, daily journal, and task management fully. The Task app for mobile only arrived January 2025. Whiteboard navigation and card arrangement on mobile work, but manipulating complex spatial layouts on a small screen is awkward. Users who want to review and add quick notes on mobile are well-served; users who want to do their primary research synthesis on a phone or tablet are not.

The lack of extensibility frustrates power users coming from Obsidian. There are no community plugins, no custom CSS, no API beyond MCP, and no way to build Dataview-style computed databases. Heptabase's tag and relation system is functional but not as programmable as Obsidian's plugin ecosystem or Tana's supertags.

Finally, the pricing structure has no free tier after the 7-day trial. For a tool in a category where Obsidian is free and Notion offers a generous free plan, the hard paywall creates a discovery barrier. The 100 AI credit ceiling on the Pro plan also means many users who want full AI functionality effectively need the Premium plan at $17.99/mo, not the base $8.99.

Who Heptabase is for

The right fit: Heptabase is best for solo learners and researchers who need to ingest many types of source material (PDFs, videos, articles, notes) and synthesize them into a coherent spatial argument. PhD students, journalists doing source-heavy investigations, self-directed learners working through technical domains, and writers doing research-intensive long-form projects are the stated and observed primary users. If your workflow involves highlighting PDFs, watching video lectures, and then trying to see how the pieces connect, Heptabase's built-in, AI-assisted pipeline is meaningfully faster than assembling the same workflow from Obsidian plugins.

Skip this when: You need team collaboration at scale, multi-user access with role permissions, or real-time co-editing beyond the single whiteboard sharing added in May 2025. You need a project management tool with tasks, assignments, and timelines; Heptabase has a basic task system but is not a project manager. You need local file storage, version-controlled markdown notes, or plugin extensibility. You are unwilling to pay from day 1 with no free tier to evaluate over months. You are primarily a mobile user expecting a full desktop-equivalent experience on iOS or Android.