Skip to content

Notes your AI can read and write

Short answer

Most notes just sit there. Hjarni is a knowledge base with a built-in MCP server, so ChatGPT and Claude read, search, and write your notes directly. Save something once and both assistants remember it, in every conversation.

Give your AI a memory

Free to start. No credit card.

The problem with normal notes

Your notes are in one app. Your AI is in another. So every conversation starts from zero. You paste in the same background, re-explain the same preferences, and copy the same context out of an old chat. Nothing you told it yesterday carries over.

The fix is not a better note app. It is notes your assistant can actually reach.

How it works

Two steps. Save your notes in Hjarni, then connect it to ChatGPT or Claude with the built-in MCP server. From then on your notes are live context: your AI searches them, reads them, and writes new ones back, without you pasting anything.

See it work

You, in ChatGPT

"Save a note: the family reunion is in Porto, first weekend of August."

ChatGPT

"Saved to your notes."

You, a week later, in Claude

"Help me plan the reunion weekend."

Claude

"Porto, first weekend of August. Here is a plan for those two days."

One note. Written by ChatGPT, read by Claude. That is the point: the notes are the shared memory, not the chat.

Simple on purpose

Hjarni keeps the shape small: notes, folders, tags. That is enough structure for a knowledge base ChatGPT and Claude can read, search, and write without guessing where things belong.

What is left out matters. No databases, no kanban boards, no page builders. Those are useful in other products. They make a knowledge base heavier than it needs to be.

Your notes stay plain Markdown you own and can export. The point is not to model everything. The point is to keep durable context in a form both people and AI can understand.

Built-in AI memory is not enough

Built-in memory is useful for small preferences. A knowledge base is for context you want to keep, inspect, edit, and share across ChatGPT and Claude.

Built-in memory Knowledge base you own
Usually stays inside one assistant. ChatGPT and Claude do not share the same memory by default. Portable across assistants. ChatGPT and Claude read the same notes through the built-in MCP server.
You can view some saved items, but it is not a Markdown knowledge base you edit like normal documents. You can open, edit, rename, tag, and file the memory yourself.
A memory saved in one assistant does not automatically survive when you work in the other. The note survives across both ChatGPT and Claude because it lives outside the chat.
Export is limited and product-specific. Export your knowledge base as Markdown when you want to leave.

Use both. Let built-in memory keep light preferences. Put the notes worth keeping in Hjarni. See Hjarni vs ChatGPT Memory for the longer comparison.

What you keep in your notes

Keep the context you would otherwise repeat. The small facts. The current plans. The decisions that explain why things are the way they are.

Preferences

How you like answers written, what you avoid, what matters when options are close.

Trip details

Dates, bookings, places to revisit, dietary notes, and the kind of pace you prefer.

People

Who someone is, what you last discussed, and details you want to remember next time.

Ongoing projects

Goals, open questions, constraints, next steps, and useful background.

Decisions

What you chose, why you chose it, and what would make you revisit it.

Reference notes

Ideas, quotes, checklists, plans, and anything ChatGPT and Claude should be able to find later.

Start with one note

You do not have to import anything or set up folders first. Connect Hjarni, then tell your assistant one thing worth remembering. Paste this into ChatGPT or Claude once you are connected:

Save a note in Hjarni about who I am and how I like you to work with me, so you remember it next time.

That first note is the whole idea in miniature. Your AI writes it, files it, and reads it back in the next conversation. From there you add notes, folders, and tags as you go. It is plain Markdown you own and can export anytime.

Common questions

Common questions

What are notes ChatGPT and Claude can read and write?

Notes stored in a knowledge base with a built-in MCP server. Once you connect it, ChatGPT and Claude can search your notes, read them, create new ones, and update existing ones during a conversation. Hjarni is one: plain Markdown notes, folders, and tags, with the MCP server built in.

How is this different from Apple Notes or Google Keep?

Those store text for you to read. They have no way for an AI assistant to reach in. Hjarni adds a server that ChatGPT and Claude connect to, so your notes become memory your AI uses on every conversation instead of context you paste in by hand.

Does ChatGPT remember my notes between chats?

Yes. Your notes live in Hjarni, not in a single chat. Save something once and both ChatGPT and Claude read it in every future conversation. Your AI stops starting from zero each time.

Do I have to organize my notes myself?

No. You can tell ChatGPT or Claude to save a note and it files it for you. Set folder-level instructions once and your AI follows your conventions: the right folder, the right tags, the format you want.

Is it free to start?

Yes. The free tier includes full MCP access with no token limits and no credit card. Save notes, connect ChatGPT and Claude, and see if it fits before you pay for anything.

Start here

Write once. You both remember.

Free to start. No credit card required.

Works with Claude and ChatGPT today.