Until now, work in Multiply stayed inside its own project. If you solved something in one project, a task in another project had no way to know about it. You'd end up re-explaining things that already existed somewhere else.
Connected Work fixes that. It's not a new feature so much as a change to how every task gets its context. A task can now pull in what it needs from its org, its workspace, its own project, and any other project or task in that workspace. If the answer already exists somewhere in your workspace, the task can find it.
References work differently too. Before, a reference pointed at a whole document, and you had to go read the whole thing to find the part that mattered. Now it can point straight at the relevant section.
We've also given the model a better sense of where it is. It sees the org, workspace, project, and task structure as a tree, with its own position marked, similar to a "you are here" on a map. That's what lets it tell the difference between context that's close and directly relevant, and context that's further away and less important.
In practice, this means less repeating yourself across projects, less digging through documents for the one part you need, and answers that point to the specific source instead of a whole pile of it.
It's a small piece of a bigger goal: a system that understands how your organization is actually structured, instead of treating every task like it's working alone.