Branches Are Where You Think. Documents Are What You Keep.
Picture the end of a good research session in KnowTree. You started with one question. It grew into a tree: one branch stress-tested the science, another chased the counterarguments, a third asked a different model to poke holes in the first two. Each branch became its own little idea space — focused, uncontaminated, free to go deep.
And then what? All of that thinking lives inside a chat interface. You can't hand a conversation tree to your manager. You can't paste it into your notes, attach it to a meeting invite, or commit it to a repo. The research was real work — and the moment you close the tab, it stops working for you.
That's the gap we just closed. KnowTree now turns any part of your tree — one branch, or several merged together — into a markdown document you can preview, download, and take anywhere.
Why Markdown, Specifically
We didn't pick .md out of habit. If your document is ever going back into an AI model — and in 2026, almost every document is — markdown is simply the cheapest, most reliable format there is.
Models speak it natively. LLMs were trained on oceans of markdown — README files, documentation, forum posts. When you feed markdown back in, nothing is lost in translation: the structure is the text.
It's the cheapest format you can buy tokens for. A PDF or Word file has to be parsed or converted before a model can read it, and the format itself carries layout machinery that bloats the token count without adding a single idea. Markdown is plain text. A heading costs two characters — and you pay the token cost every single time the document re-enters a conversation.
It goes everywhere. Notion, Obsidian, GitHub, Google Docs, your team wiki — everything imports markdown. It's readable raw, it diffs cleanly in version control, and it will still open in thirty years.
From a Branch to a Document in One Prompt
Next to the search and citation controls in the message box there's now a document button. Arm it, describe what you want — "a one-page brief for my physician", "a decision memo with the trade-offs we found" — and send. The model replies in the chat with a short summary, and the full document arrives as a file card: title, type, download. Click the card and the document opens in a preview panel right where the Knowledge Map usually sits.
Merging Idea Spaces
Here's where the tree structure pays off. The whole point of branching is that different lines of thought develop in isolation. But a real conclusion usually needs all of them.
Choose Merge branches and the Knowledge Map becomes a selector. Your current path is pre-marked; tap any node in any other branch and that branch lights up. A live counter shows how many branches and nodes of context you've gathered. Then just type your prompt — the document is written with all the selected idea spaces in view.
Two details matter. First, overlap is handled for you: if two selected branches share part of their history, the shared context is sent once, not twice. Second, the merge is drawn on the map as dashed lines from every source branch into the new node, permanently — a faithful record of exactly which context every document had access to. From that point on, the conversation continues with the merged context.
Research That Outlives the Session
Explore wide, in parallel, with whichever models suit each branch — then converge everything into a brief, a decision memo, a chapter for your knowledge base. The exploration stays in the tree, where it's navigable. The conclusion leaves as a file, where it's useful.
Documents are live for everyone, on every plan. KnowTree has a free tier — no API keys, no setup.