Capacities

Capacities is a personal knowledge management app built around a typed-object data model. Founded in 2022 by Steffen Bleher and Michael von Hohnhorst in Germany, the company is entirely bootstrapped with no venture capital. Every piece of content in Capacities is an "object" with a type: a Book, a Person, a Project, a Meeting. Each type carries its own schema of properties. Notes link to objects, objects link to each other, and the whole structure is queryable. The result is closer to a personal database than a document editor, but without the configuration overhead of building one from scratch.

The app runs on Web, Mac, Windows, Linux, iOS, and Android, with offline-first sync added in January 2025. The Pro tier ($9.99/month) unlocks the AI assistant, smart semantic queries, calendar integration, task actions, and Readwise/Kindle sync. The free tier is substantial: unlimited notes, 5 GB storage, daily notes, graph view, and native mobile apps. As of April 2026, the team ships roughly two releases per month. Search 2.0 launched in January 2026, AI image analysis in April 2026, and bulk import entered Believer beta the same month.

What Capacities actually is in April 2026

The "object" concept is what separates Capacities from every other notes app. In Notion, you have pages and embedded databases. In Obsidian, you have markdown files and plugins. In Capacities, the atomic unit is a typed object. When you create a note about a book you're reading, you don't drop it in a Books folder or tag it #book. You create a Book object. It has properties (author, status, your rating), it links back to any Person objects you mentioned, and it surfaces in every query that filters by type:Book.

Built-in object types include Pages, Daily Notes, Images, Web Links, PDFs, and Tasks. Custom types are where the system opens up: researchers build Paper and Experiment types, consultants build Client and Meeting types, writers build Character and Chapter types. Each custom type carries whatever properties you define. The graph view shows how objects connect. Smart queries (Pro) let you ask the database things like "show me all Books tagged philosophy, linked to people I met this year."

The team shipped 45+ releases since 2022. In 2025-2026 alone: offline-first architecture (January 2025), Kanban view (October 2025), Readwise integration (November 2025), task management (December 2025), Search 2.0 with exact matching and offline search (January 2026), AI image analysis transforming photos into searchable objects (April 2026), and AI Chat Connectors enabling ChatGPT, Claude, and Cursor to read and write your Capacities data via MCP (April 2026). The roadmap lists Agentic Chat as the next major feature.

"Capacities has really hit it out of the park with solid fundamentals and a unique approach to knowledge management." - Parth Shah, Android Police, 2025

Where Capacities sits versus Tana and Notion

The PKM space in 2026 has three dominant architectures. Tana uses node-based outlining, where every bullet is a "node" that can be tagged via "supertags" into structured objects. Notion uses the block-document model, with pages made of blocks, with relational databases embedded separately. Capacities uses typed-object databases as the primary layer, with documents hanging off them rather than the other way around.

Capacities vs. Tana: Tana's supertag system is more granular. You can turn any inline text into a structured data point by tagging it. It suits users who want to work in outlines and build complex automation via Tana's command line. The tradeoff: Tana is web-only, has no native mobile app, no Readwise integration, and has a steeper configuration curve. Capacities wins on accessibility (native iOS/Android, offline sync), breadth of integrations (Readwise, Kindle, Raycast, WhatsApp, Telegram), and calendar-linked daily notes. Tana wins on outlining depth and automation for power workflows.

Capacities vs. Notion: Notion is designed for teams: real-time collaboration, shared workspaces, permission management, 50+ view types including Gantt. Capacities has none of that. It is single-user only. On the other hand, Notion's mobile apps are web wrappers with limited offline access. Capacities ships native mobile apps and full offline-first sync. The data model difference is significant: in Notion you configure a database and embed it in a page; in Capacities the typed object exists independently and pages reference it. Capacities is faster to set up for personal PKM; Notion is the better choice the moment two people need to work in the same space.

"The right amount of Obsidian + Roam + Notes + Notion." - Jon Soini, Product Hunt, 2022

How AI actually works inside Capacities

AI in Capacities is first-party, embedded in the data model, and gated behind Pro. Three distinct layers exist as of April 2026:

AI Assistant: Available inline inside any object or daily note. Write a prompt, the assistant reads context from surrounding content and your knowledge base, and returns text you can edit or insert. Use cases: summarizing a book note, drafting a follow-up from a meeting object, expanding a fleeting idea. You can configure which AI provider powers it. In November 2025, Capacities added support for custom AI model providers, and Perplexity was added as an option alongside OpenAI and Anthropic.

Smart Queries: This is where AI integrates with the object model directly. Smart queries use semantic search across your entire knowledge base. Type "notes about machine learning" and it surfaces content tagged neural networks, LLMs, or gradient descent, not just notes containing the literal phrase. Variable queries let you build templates with embedded smart queries that auto-populate. This is the feature Tana's power users envy most.

AI Chat Connectors (Beta, April 2026): The newest layer: Capacities connects via MCP to ChatGPT, Claude, Cursor, and Le Chat. Your external AI assistant can now search your notes, read specific objects, save to daily notes, and add links, bridging the gap between Capacities as a knowledge store and AI assistants as reasoning engines over that knowledge.

AI Image Analysis (April 2026): Photos added to Capacities are analyzed by AI and converted into searchable, described objects. A photo of a whiteboard becomes findable by its content. This extends the object model to visual media.

The friction users keep raising

Bulk import is still incomplete. Moving an existing Notion workspace or Obsidian vault into Capacities requires mostly manual work. Bulk import entered Believer beta in April 2026. It is not yet generally available. This is the most-cited reason users don't fully commit to a switch.

Mobile bugs persist. Multiple reviewers note iOS text editing issues: text disappearing on tap-to-edit, non-functional UI elements, and feature parity gaps between iPhone and iPad. Functions available on iPhone are sometimes missing on iPad and vice versa. The desktop and web experience is consistently praised; mobile lags behind.

No collaboration. Capacities is built for one person. There are no shared spaces, no real-time co-editing, no comment threads, no permission levels. This is a design choice, not a roadmap gap. The team has stated the focus is the individual knowledge worker. But for anyone who needs to share a research wiki with a colleague, Capacities is not the answer.

AI locked to paid tier. Smart queries, the AI assistant, and AI image analysis all require Pro ($9.99/month). The free tier is genuinely useful but shows you the interface for features you cannot access, which some users find frustrating.

Learning curve. The object model clicks for some users immediately and confuses others for weeks. Transitioning from a folder-based mental model requires rethinking what a "note" even is. New users often start by using Capacities like a traditional notebook and only discover the object layer after reading documentation.

"That learning curve is REAL. Not for casual note-takers." - Alesia Cherniavskaia, Product Hunt, 2024

Who Capacities is for

Best fit: Solo researchers, writers, students, consultants, and knowledge workers who consume a high volume of information (books, articles, meetings, people) and need to connect and retrieve it later. The typed-object model rewards people who think in categories, specifically those who naturally want to distinguish a Book from an Article from a Person from a Meeting. Daily notes plus mobile capture is a strong combo for anyone who journals or logs throughout the day. Readwise users who want their highlights integrated into a queryable knowledge graph will find Capacities a natural home.

Skip this when: You need to share a workspace with anyone else. You have thousands of existing notes and need bulk import now, not in beta. You want Gantt charts, sprint tracking, or database views beyond Kanban. You need AI features but are not willing to pay $9.99/month. You are a casual note-taker who just wants a fast scratchpad. The app's richness will feel like overhead.

Annette Raffan, a PhD researcher who switched from Obsidian: "Images are 'Objects' in their own right." She cited this as a key reason Capacities fit her research workflow better than Obsidian's file-based model, where images were inert attachments rather than queryable entities.