Nodes & edges
How knowledge is represented: nodes, the edges that connect them, and the hierarchy they form.
Everything in a Naumu space is built from two pieces:
- Nodes are the things you care about - a person, a feature, a meeting, a decision.
- Edges are the relationships between them.
Together they form your memory: a connected map of your knowledge that you and your AI agents can read, search, and grow.
Nodes
A node is a single unit of knowledge. Every node has:
- A label - its name, shown everywhere it appears (for example, "Onboarding flow").
- A type - the kind of thing it is (Feature, Task, Person, Meeting…). Types come from your space's schema - the set of rules that defines what kinds of things and relationships your space allows. A node's type also gives it a color throughout the app.
- Content - an optional body of text about the node, written in a rich-text editor.
- Attributes - structured fields defined by the node's type, such as a status, a priority, a date, or a number. Which fields are available depends on the schema for that type.
Naumu also records system metadata you don't edit directly, including when a node was created and last updated.
Most nodes are created for you. As you chat in a space, Naumu picks out the things worth remembering and adds them as nodes, wired into your existing memory. You can also work with nodes by hand from the node detail panel - rename one, write in its body, or fill in its attributes.
Let chat do the heavy lifting. Describe what happened in plain language and Naumu turns it into nodes and edges for you. See Quickstart.
Edges
An edge is a connection between two nodes. It has a direction - a source and a target - and a relation: the label that says what the connection means, such as Resolves, Relates to, or Owns.
Edges fall into two groups:
- Parent edges define the hierarchy. A parent edge runs from a child up to its parent - so a Task can sit under the Feature it belongs to.
- Other connections are everything else: side-to-side, many-to-many links that relate nodes without nesting one under the other. Relates to is the general-purpose fallback when no more specific relation fits.
Your schema suggests the relations that make sense for a given node - it knows, for example, that a Bug Resolves a Feature. But you're never limited to those suggestions: Relates to always remains an option.
Hierarchy & breadcrumbs
Parent edges - and only parent edges - form your space's tree. Follow a node's parents upward and you reach its ancestors; follow its children downward and you reach its descendants. Side-to-side connections are never treated as hierarchy, so they don't affect a node's place in the tree.
This tree is what powers breadcrumbs: the trail that lets you move from a broad topic down to a specific detail and back up again. A well-shaped tree keeps a space easy to read as it grows.
Reparenting
Reparenting means moving a node - along with everything beneath it - to a new parent. Since the hierarchy is built only from parent edges, this just re-points the node's parent edge; its side-to-side connections and attributes stay intact.
Reparenting is how you reorganize a space as your understanding evolves - pull a misfiled Task under the right Feature, or promote a node to sit higher in the tree.
The node detail panel
Opening a node brings up its detail panel - one place to see and shape everything about it. Every node uses the same layout, top to bottom: its type and title, a conversations card, its properties, its files, and its body. From here you can:
- Edit the title inline.
- See the latest conversation about the node on the card at the top; N more chats opens every conversation, split into Direct, Related, and Sources.
- Edit the node's properties in place - its attributes, its Parent, and Created / Updated. Drag a property to reorder it, or open its label's menu (click or right-click it) to Hide attribute; hidden ones collapse into an N more attributes row. Order and visibility apply to every node of that type. New property adds an attribute to the type without leaving the panel. A warning icon at the end of a row flags a value that breaks the schema.
- Browse its Files - a single grid that gathers the notes and uploaded files tied to the node. Note cards carry an Origin badge when the note first created the node, or a Modified badge when it later changed it. The grid only appears when there is something in it.
- Write in its body, a rich-text editor that teammates can edit at the same time, with
@to mention a person or node and#to tag a topic. - Open Context & connections from the icons in the panel header for an AI-written context summary and the node's connections, grouped by node type or by edge type, with Add connection. The History icon next to it opens the node's edit history.
- Ask Naumu a question about the node directly from the panel.
The ⋯ menu holds Show memory, which opens the memory graph zoomed onto the node, Visibility, Copy Link, and Delete.
Deleting
Deleting a node removes it from your memory. Because the hierarchy is built from parent edges, deleting a node also deletes its descendants - every child, grandchild, and so on beneath it.
When a node has descendants, Naumu asks you to confirm first and tells you exactly what will go (for example, "This will also delete 2 Tasks and 1 User Story"). Deletion cannot be undone, so read the summary before confirming.
Deleting a node cascades down its parent edges. A high-level node can take a large subtree with it - read the confirmation summary before you delete.
Related
- Spaces - how nodes and edges live inside a space.
- Schema - the node types and relationships that give your memory structure.
- Quickstart - build your first memory from a conversation.