Browse documentation

Create a data object

Create your first table-like data object in a workspace, and optionally populate it from a CSV.

A data object is a structured collection of records, like a table with typed columns. It's the foundation everything else in Spark builds on, so creating your first one is an early step. This page covers creating an object and, optionally, populating it from a CSV as you go.

Create a data object

Data objects are created from the admin view, so start there.

  1. In the sidebar (Home tab), click Data.
  2. Click Manage at the top of the data view. This opens the Data Objects admin page, the only place data objects are created.
  3. Click New, then Create from scratch.
  4. Fill in the New Data Object dialog:
    • Icon: an optional symbol shown next to the object.
    • Name: a clear, human-readable name (for example, Counterparties).
    • Identifier / Slug: the stable, machine-friendly id used in APIs and references. It is filled in from the name, and you rarely need to change it.
    • Group: which section the object appears under, defaulting to Main. See Organize objects into groups below.

You can create the object empty and add records later, or import a CSV right here as you create it.

The New Data Object dialog with Icon, Name, Identifier / Slug, Group and an optional CSV drop zone for attributes.
The New Data Object dialog: name it, pick a group, and optionally drop a CSV to build the attributes.

Populate it from a CSV as you create it

The New Data Object dialog has a CSV drop zone, which is the quickest way to go from a spreadsheet to a populated object:

  1. Drag your CSV onto Drop your CSV here (or click to browse).
  2. Click Extract attributes with AI. Spark's AI model reads your columns and proposes the object's attributes, inferring a sensible type for each, which you can review and adjust.
  3. Leave Ingest file's content checked to import the rows along with the schema.
  4. Click Create.
After a CSV upload, a File uploaded successfully message and an Extract attributes with AI button.
Once the CSV is uploaded, click Extract attributes with AI.

Spark proposes one attribute per column with an inferred type. Untick any column you don't want, rename it, or change its type from the dropdown before you create the object.

The AI-proposed attributes list, each row with a checkbox, a name, and a type dropdown such as string or date.
Review the proposed attributes: keep, rename, or retype each before creating.

Every row becomes a record you can work with immediately.

Tidy the source before you import: one row per record, a header row with clear column names, and consistent date and number formats. A few minutes cleaning the CSV saves more time than fixing types afterward.

Configure the object

Once the object exists, open it from the Data Objects admin list to configure it. The editor has four tabs:

  • General holds the object's Icon, Name, Identifier / Slug (read-only), Description, and Group. Changes save as you make them, so there's no separate save button.
  • Attributes is the object's columns. Click Add Attribute to add one, click a row to edit it, and use a row's ⋮ menu to copy its slug, open display options, or Archive it (archiving is a soft delete; the data stays safe). Each attribute shows its type, description, properties, and how many validators it has. See Attribute types for what each type does and how to configure relationships, formulas, and the rest.
  • Reference Display controls how a record of this object looks when another object links to it. Pick up to 5 text, number, or date attributes, drag them into the order you want, and set a Separator to join them; a live Preview shows the resulting label.
  • Constraints sets uniqueness rules. Add a constraint with one attribute to make that column unique, or several attributes to require the combination to be unique. The system UUID primary key is always unique regardless of what you set here.
The data object editor with the four tabs, showing the Reference Display tab: available and selected attributes, a separator, and a live preview of the reference label.
The object editor's four tabs, here on Reference Display: pick and order attributes, set a separator, and preview the label.

Spark infers a type for each attribute when it reads your CSV, but you can change any of them before creating the object, and add or adjust attributes later on the Attributes tab. The type is locked once an attribute exists, so it's worth getting right early.

Other ways to bring in data

Creating an object with a CSV is one path. Depending on where your data lives and how often it changes, you might use another:

  • Into an object you already have: import a CSV from its record grid, with AI-assisted column mapping. See Add, edit & import records.
  • On a recurring schedule: pull files automatically from a provider with Set up SFTP.
  • In response to an event: fetch and load data with a workflow.

Organize objects into groups

As a workspace fills with data objects, you can sort them into named, collapsible sections. Grouping is done from the admin view where objects are created (Data, then Manage).

  • Create a group: at the bottom of the object list, click New group, type a name (up to 30 characters), and confirm with the check button or Enter.
Creating a new group inline at the bottom of the object list, with confirm and cancel buttons.
Creating a group: type a name and confirm with the check.
  • Put an object in a group: open the object's ⋮ menu and choose Move to group, then pick a group, or Main to take it out of any group. Objects that aren't grouped sit in a section called Main.
  • Rename or reorder a group: use the group header's ⋮ menu (Group actions) for Rename, Move up, and Move down.
  • Delete a group: the same menu's Delete removes only the group. Its objects move back to Main and no data is deleted.

Collapse or expand a section with the chevron on its header, or use the Collapse all and Expand all toggle next to the search box. On the regular Data page these sections are read-only; grouping is managed from the admin view.

Where to go next