Skip to content

Hjarni vs Roam Research

Roam is a networked outliner with block references. Hjarni is a Markdown knowledge base with an official, built-in MCP server.

Free to start. No credit card required.

Quick answer

Roam Research is a networked outliner for power users: daily notes, block-level references, wiki links, and queries over a graph. It has no official MCP; community servers built on Roam's graph API can read and write your graph, but they are not vendor-backed and you set them up yourself. Hjarni is a Markdown knowledge base with an official, hosted MCP that Claude, ChatGPT, and other MCP clients reach out of the box. Pick Roam for block references and graph thinking. Pick Hjarni for an official MCP, plain Markdown, and a free tier to start.

Hjarni Roam Research
Official MCP server

Hjarni's MCP is official and built in. Roam has no official MCP; community servers use Roam's graph API and are not vendor-backed.

Built-in
Read and write through MCP

Roam's community servers can read and write via its backend graph API with a token. Hjarni's read and write MCP is official and built in.

Built-in Community
Hosted, remote MCP for cloud AI

Hjarni answers MCP calls at a hosted endpoint. Roam relies on community tooling you set up yourself.

Built-in
Block references and graph queries

Roam's block-level references, daily notes, and queries are its core strength.

Built-in
Plain Markdown notes

Hjarni stores plain Markdown. Roam uses a block and graph model and exports Markdown.

Partial
Export to Markdown

Both export Markdown. Roam exports per-page and full-graph; Hjarni exports a Markdown ZIP.

Built-in (ZIP)
Permanent free tier

Hjarni has a free tier (25 notes). Roam offers a 31-day trial, then paid only.

Starting price

Roam has no permanent free plan; Pro is 15 USD per month.

Free, Pro from 9 EUR/mo Pro 15 USD/mo
Full-text search

A networked outliner versus an AI-native note store

Roam Research is a tool for thought built around a graph. Daily notes, block-level references, wiki links, and queries let power users connect ideas at a fine grain. If you like thinking in linked blocks and assembling a personal graph, Roam is a deep, well-loved option.

Hjarni is built around a different goal: giving your AI a memory. It is a knowledge base of plain Markdown notes that assistants read and write through an official MCP server. The aim is durable, AI-readable context, not block-level graph mechanics.

Official versus community MCP

This is the distinction worth being precise about. Roam has no official, vendor-built MCP server. There are good community and open-source servers that connect to Roam's official backend graph API with a token, and they can read and write: creating and editing pages and blocks, and searching. But they are community-maintained, and you set them up and trust them yourself.

Hjarni's MCP server is part of the product. It is official, built in, and uses OAuth, so connecting Claude or ChatGPT does not depend on third-party tooling. If you want the integration supported by the vendor rather than the community, that is the difference.

If you want block references and a graph to think in, Roam is excellent. If you want an official MCP and plain Markdown, Hjarni is the simpler path.

A concrete workflow difference

In Roam, you might keep a running graph: daily notes that link back to project pages, block references that surface a thought wherever it is relevant, and queries that pull related blocks together. It rewards careful, ongoing linking.

In Hjarni, you keep notes in folders and connect an assistant. Ask Claude to find recurring themes across a folder, draft a summary, and write it back as a new note. The structure is simpler; the payoff is that any AI you connect works from the same shared notes.

When Roam is the better fit

If you want a heavy-duty networked outliner with block references, a daily-notes workflow, and graph queries for personal thinking, Roam is a strong choice. It is built for exactly that kind of deep, linked note-taking.

When teams pick Hjarni instead

The case for Hjarni is an official, hosted MCP without community setup, plain Markdown you own, and a free tier to start. Roam has no permanent free plan, so Hjarni is also the lower-friction way to begin if you mainly want notes any AI can read and update.

Hjarni notes are plain Markdown, exportable as a ZIP anytime, hosted in the EU.

When to use Roam Research

  • You want a heavy-duty networked outliner with block references
  • You want a daily-notes workflow and graph queries
  • You are a power user who thinks in linked blocks

When to use Hjarni

  • You want an official, hosted MCP without community setup
  • You want plain Markdown and a free tier to start
  • You want a simpler knowledge base any AI reads and writes

Roam links your thoughts into a graph. Hjarni gives your notes an official memory any AI can reach.

Common questions

Common questions

What is Roam Research?

A networked note-taking app and tool for thought, built around a block-level outliner with daily notes, bidirectional wiki-links, block references, and queries.

Does Roam have an official MCP server?

No. Roam has community-built MCP servers that use its backend graph API with a token, supporting read and write. Hjarni's MCP server is official and built in, and uses OAuth.

Can I import a Roam graph into Hjarni?

Yes. Roam exports per-page and full-graph Markdown. Import the export into Hjarni and folders and links come across.

When should I pick Roam over Hjarni?

Pick Roam when you want a heavy-duty networked outliner with block references, a daily-notes workflow, and graph queries for personal thinking. Pick Hjarni when you want documents any AI reads and writes through a built-in, hosted MCP server.

Does Roam have a free plan?

Roam offers a 31-day trial, then paid plans (a monthly Pro subscription, plus a longer-term Believer plan). Hjarni has a free tier with full MCP access up to 25 notes.

Give your AI a memory

Write once. You both remember.

Free to start. No credit card required.

Works with Claude and ChatGPT today.