These two end up on the same shortlist because they promise the same outcome — notes that connect rather than pile up — and disagree completely about how you get there. Capacities says structure should be built into the note. Obsidian says structure should emerge from links between plain files. (Disclosure: we make a third app, Clair Mind, which does neither. Referee only.)
At a glance
| Capacities | Obsidian | |
|---|---|---|
| Core unit | A typed object: book, person, meeting, idea | A markdown file in a folder |
| Structure | Declared up front — each type has properties | Emerges from [[links]] and folders |
| Storage | Its own cloud, with offline support | Plain .md files on your disk |
| Price | Free core product; Pro around $12/month, cheaper billed yearly | Free for personal use; Sync and Publish are optional add-ons |
| Extensibility | Limited, by design | Very large plugin ecosystem |
| Mobile | Good, native | Good, and improved a lot |
| Learning curve | Low to start, medium once you design types | Medium, higher with plugins |
| Lock-in risk | Real — the structure is the value and it lives there | Low — the files are already yours |
Prices change and both sites are the authority; check before you commit.
What Capacities is actually doing
The idea is that most notes are secretly about a kind of thing. A note about a book has an author, a status and a rating. A note about a person has a company and a last-contacted date. A note about a meeting has attendees. Capacities makes those types first-class: you define object types, give them properties, and every note you create is an instance of one.
The payoff is real and shows up quickly. Because the app knows Ali is a person and Thinking, Fast and Slow is a book, the connections between them are typed too, and every object page automatically collects everything that references it. You get a genuinely useful "here is everything about this client" page without maintaining an index by hand, which is what maps of content in a file-based system are trying to approximate with effort.
It also has a daily note at the centre, so capture goes into one stream and gets attached to objects as you go. The Pro tier adds an AI assistant and calendar integration.
The cost is that you are now designing a schema. Deciding what counts as a type, and which properties it needs, is a modelling exercise, and modelling before you have the data is how people end up with eleven object types and four notes. It rewards people who already know the shape of their work and punishes people who are still finding out.
What Obsidian is actually doing
Obsidian's position is that structure imposed early is usually wrong, so it ships almost none. You get markdown files, [[wikilinks]], backlinks and a graph, and everything else is a plugin or a convention you invent.
The files are the point. They sit on your disk, readable by any editor, backupable, greppable, version-controllable, and readable in twenty years by software nobody has written yet. This is the strongest available answer to the question every notes app eventually raises, which is what happens when the company changes its mind — see plain text notes and export your notes.
Obsidian can reach a fair distance toward the Capacities model with properties in frontmatter and the Dataview or Bases-style query plugins: a table of every book with a rating, a list of everyone you have not contacted in a month. It works. It is also assembly — you are building the feature Capacities ships.
Pricing is unusual and worth stating plainly: the app is free for personal use, Sync and Publish are separate optional subscriptions, and commercial use has its own licence.
The honest trade
Capacities' cost is where the value lives. Your text will export. The type system, the properties and the automatic relations are what you built, and those are the hardest part to take anywhere else. On a subscription that is a compounding commitment, so ask the export question in month one rather than year three.
Obsidian's cost is that it never tells you what to do. A meaningful number of people spend three weekends choosing plugins and folder schemes and never get to the notes. That failure is common enough to be the headline entry in PKM mistakes.
Both assume the hard part is organising what you have collected. For many people the hard part is earlier — the thought never reaches any app at all.
Choose Capacities if
- Your notes are genuinely about recurring kinds of things: clients, papers, recipes, meetings, films.
- You want backlinks and collections to happen without maintaining them.
- You have tried a file-based vault and bounced off the amount of setup.
- You prefer paying a subscription over spending evenings configuring.
Choose Obsidian if
- You want your notes to be files you own, indefinitely.
- Your material is mostly prose and links rather than records with fields.
- You are happy to assemble your own setup, and you might want plugins.
- You want it free, or you are keeping a decade-scale Zettelkasten.
The short verdict: if you can name the object types in your life without straining, Capacities will feel like the app finally understood you. If you cannot, the schema becomes a chore and Obsidian's refusal to guess ages better. For a Zettelkasten, Obsidian remains the default; for a structured personal database with notes attached, Capacities is doing something Obsidian only imitates with plugins.
Neither fixes capture
Both are desktop-shaped tools for deliberate work at a desk. If your real bottleneck is that ideas arrive while you are walking and never get written down, a better vault schema will not help, because nothing is going in. That is the problem Clair Mind is built for: one-tap capture on the phone, offline and private, with connections drawn for you. Different bottleneck, different tool.
Related: Notion vs Obsidian, Obsidian vs Logseq, and the wider note app comparisons.