Most note-taking apps treat notes like files in a cabinet: you write something down, file it away, and hope you remember where you put it. That model works fine for storage. It fails completely for thinking.
Ideas don't live in isolation. The note you wrote about a customer call last month is connected to the product spec you drafted yesterday, which is connected to the strategy memo you'll write next week. A filing cabinet can't represent those relationships. A graph can.
Two brackets, zero ceremony
In NoteLounge, typing [[ anywhere in a note opens a quick picker over every note you've written. Choose one — or type a title that doesn't exist yet — and you've created a link. That's the whole feature. No dialogs, no link manager, no leaving the keyboard.
The low ceremony is the point. Linking has to be cheaper than not linking, or you simply won't do it mid-thought. We measured ourselves against one test: can you connect two ideas without losing the sentence you were writing? If the answer is ever no, the feature has failed.
Backlinks do the remembering
Every link is bidirectional. When you open a note, NoteLounge shows you every other note that references it — even ones you wrote months ago and forgot about. You don't have to maintain an index, because the index maintains itself.
This is where the compounding happens. A note you write today gets more valuable every time a future note links to it. After a few months, opening an old note feels less like reading an archive and more like rejoining a conversation.
Links are how the agent thinks, too
Wiki-links aren't just for you. When you ask the Lounge agent a question, it traverses the same graph you built — following links from relevant notes to their neighbors to assemble context a keyword search would miss. The structure you create while writing becomes the structure the agent reasons over.
That's the quiet bet behind NoteLounge: a knowledge base you build for yourself, one [[ at a time, turns out to be exactly the substrate an AI needs to be genuinely useful. You're not doing extra work to help the machine. You're thinking in connections, and the machine gets to come along.