Spaces overview
A space is a container that groups records inside a module. Each module can hold an unlimited number of spaces and subspaces, so it stays organized and clear even with a large amount of data.
Where spaces sit in the data model
| Level | What it is | Example |
|---|---|---|
| Module | A type of data — the table. | Deals |
| Space | A container that groups records within that module. | EU pipeline |
| Record | A single row of data. | One deal |
See Hierarchy – modules, spaces and records for the full picture.
Why use spaces
- Keep a module readable — split its data into logical groups: by team, client, region, or dataset.
- Control access per space — only the creator and invited users can open a space; what each can do is set by their space role.
- Give each space its own setup — statuses, connected field groups, a key field for deduplication, and native automations. See Space settings.
One module, one kind of data
Because a module can hold any number of spaces, it is tempting to make one module the home for everything and separate the different kinds of data into spaces — products in one, contacts in another, contracts in a third. It works, right up until someone opens the module's own view of all its records.
That view is the union of every field in every space beneath it. Spaces holding the same kind of data share their fields, so the view stays tight and readable. Spaces holding different kinds of data have almost no fields in common, so the union becomes a very wide table where nearly every cell is blank — a contact has no delivery weight, a contract has no stock level — and the one view that should give you the overview becomes the least usable screen in the system.
So give each kind of data its own module, and use spaces for variations of it. Five spaces for five product ranges is exactly what spaces are for; the records still look alike, so the module view still reads. Products, contacts, and contracts are three modules.
Small structural differences between spaces in one module are fine and normal. It is the mixing of genuinely unrelated record shapes that costs you the module view.
Spaces and agents
A space is also the unit an AI agent works in: its description tells the agent what the data holds, its key field makes agent writes deduplicate safely, and vectorization makes its records semantically searchable.
Parent and child spaces
Spaces can nest inside a module: a parent space stands on its own, a child space belongs to it — for example a parent Clients space with a child space per region. You choose this when creating a space.
