"Put this design doc into a develosaur board" is one sentence, and what comes back is not a summary of the document. It is the document's own structure, turned into a board.

What actually happens to the text

The build step behind this, grow, takes a markdown outline as its entire payload. A heading becomes a node, and how deep the heading is sets how deep the node sits in the tree. Everything else, the prose under a heading, becomes that node's content. Tags land on the leaf nodes in the same call. The whole branch attaches under whichever part of the board it belongs to, or under the root if it is a new area of work entirely. One call in, a tree out.

Why a design doc fits this so well

A design doc is already organized as headings and sections before an agent ever touches it. Turning it into a board is not asking the agent to invent a structure the document does not have; it is reading the structure that is already there and keeping it. A section on authentication becomes a branch called authentication. A subsection on token refresh becomes a task under it. Nothing gets flattened into a single list the way pasting a document into a chat usually does, and nothing gets restructured into whatever shape the agent happens to prefer that day.

It lands as work, not just as text

Tags go on in the same call that creates the nodes, so the branch is not inert structure sitting on the board waiting for someone to notice it. It is immediately visible in the ranked list the moment it lands, the mechanism covered in how the priority order decides what is next. If you say which part to start first, that becomes the pinned or claimed task; if you do not, the tree's own order decides what surfaces first, the same depth-first order the board always uses.

What you actually see

Paste in a three or four page design doc and ask an agent to put it on your board. In the next few seconds, headings become branches, the leaf-level work is tagged, and whatever you named as the priority is pinned at the top. You are not reading a paragraph describing what the document contains. You are looking at the document, restructured into something a ranked list and a lease system can actually work from, which is the same argument made in more depth in why a graph beats a list when an AI agent reads your project.

More on how the rest of the board works is on the how-it-works page, and more of these are on the blog.