Core concepts

The words Naumu uses - space, topic, thread, note, view, agent - and how memory and the model work underneath.

Updated October 8, 20264 min read

You can use Naumu without reading this page. It is here for when you want to know what a word means, or what happens under the hood.

The first six terms are the things you work with every day. The last two, memory and the model, explain how Naumu remembers.

Space

A space is one workspace. It has its own members, topics, conversations, notes, views, agents, and one memory. A common setup is one space per company, team, or project.

Switching spaces switches everything at once. See Spaces.

Topic

A topic is a channel-like area inside a space, written with a hash: #product, #sales. It does two jobs in one move: it organizes content, and it decides who can see that content. Everyone who has joined a topic can see what is filed there.

See Topics.

Thread

A thread is one conversation - the app also calls it a chat. It can be between people, with Naumu, or both. Mention @Naumu or another agent to bring it into a thread.

A thread is private, shared with everyone in the space, or filed into one or more topics. See Chatting with Naumu.

Note

A note is a rich text document in a space. Use one for anything that is easier to write as prose: meeting minutes, research, a draft. When a note is ready, Commit hands it to Naumu to save what is in it to memory.

See Notes.

View

A view is a saved, filtered list of things from memory - for example, every open task grouped by owner. You can show it as a table, timeline, calendar, or kanban board.

See Views.

Agent

An agent is an AI member of a space. Every space has Naumu built in. You can add more agents, each with its own name and instructions, or connect an AI system you run yourself.

A routine is recurring work you hand to an agent, such as a Monday digest. See Agents and Routines.

Memory

Memory is what a space knows. It has two parts:

Nodes are the individual things: a person, a task, a decision, a client. A node has a name, a type, and details such as a status or a due date.
Edges are the connections between nodes: a task belongs to a project, a feature is part of a product. Who owns or is assigned to an item is a Person field on it, not an edge - see Model.

You do not build memory by hand. Naumu adds and updates nodes as you work with it. Each time, the conversation shows a Memory updated line where you can see what changed and undo it. The full log for the space is under Explore → History. You can also edit nodes yourself.

To see all of it at once, open Explore → Model and select the Your memory card. For the details, see Nodes & edges.

Model

The model is the shape of a space's memory. It says which types of things the space keeps track of, such as Task, Client, or Decision, which details each type has, and how the types connect. It is also called the schema. You do not need to design a model first: it grows out of what you tell Naumu, and you can adjust it later.

Open it from Explore → Model, at the bottom of the sidebar, to see the types, each with the routines that keep it fresh. See Model for types, attributes, and connections, and Routines for routines.

Introduction - what you can do in Naumu.
Nodes & edges - how memory is put together.
Model - how the model is defined and edited.