Creating Tasks

Tasks can be created in a pipeline through several methods, accessed via the Create Tasks button in the Data Explorer.

CSV upload

For pipelines with dataset fields, upload a CSV file to create tasks in bulk.

1. Click Create Tasks and select the Upload tab.
2. CSV column headers must exactly match the pipeline's dataset field names. You can download a CSV template with the required headers.
3. Each row in the CSV creates one task, with the values seeded as predefined (read-only) fields.
4. Click Create.

ZIP upload

For pipelines with file upload fields, upload a ZIP file containing:

• A CSV file for structured data
• Files referenced by the CSV (paths relative to the ZIP root)

From a dataset

Seed tasks from an existing dataset in the Data Vault.

1. Click Create Tasks and select the From Dataset tab.
2. Choose a source dataset from the dropdown.
3. The platform auto-matches fields by name. Use the field mapping editor to review matches, manually map unmatched fields, or skip fields. For JSON-typed source fields, you can extract specific paths from the JSON structure.
4. Click Seed Data.

From another pipeline

Seed tasks from another pipeline's completed output.

1. Click Create Tasks and select the From Pipeline tab.
2. Choose a source pipeline from the dropdown (can be from a different project in the same organization).
3. The platform auto-matches fields by name. Use the field mapping editor to review matches, manually map unmatched fields, or skip fields. For JSON-typed source fields, you can extract specific paths from the JSON structure.
4. Click Seed Data.

Pipeline chains

Set up automatic task creation between pipelines.

1. Click Create Tasks and select the Chaining tab.
2. Choose a source pipeline from the dropdown.
3. Select a trigger type — On Task Completion (creates a task each time one completes) or On Pipeline Completion (waits for all tasks to finish, then seeds in batch).
4. The platform auto-matches fields by name. Use the field mapping editor to review matches, manually map unmatched fields, or skip fields.
5. Click Run Chain.

See Pipeline Chains for full details.

Bulk create

For pipelines without dataset fields, create multiple blank tasks at once.

1. Click Create Tasks and select the Bulk Tasks tab.
2. Enter the number of tasks to create.
3. Click Create.

Infinite seeding

Infinite seeding auto-creates tasks as contributors claim them, so one task is always available.

1. Click Create Tasks and select the Infinite Seeding tab.
2. Optionally set a target count — the maximum total tasks to create. Leave blank for unlimited.
3. Optionally select task groups under Copy values from groups to duplicate field values from existing tasks. If no groups are selected, new tasks are created with empty fields.
4. Enable infinite seeding. When a contributor claims the last available task, a new one is automatically created.

Infinite seeding is useful for:

• Data generation pipelines where contributors produce original content.
• When you want to create tasks incrementally rather than in bulk.
• Scaling up existing task data by duplicating from source task groups.

API

Create tasks programmatically using the External API.

Example

POST /api/v1/workflows/{workflow_id}/tasks
Authorization: Bearer YOUR_API_KEY
Content-Type: application/json

{
  "tasks": [
    {
      "fields": {
        "input_text": "First task input",
        "category": "example"
      }
    },
    {
      "fields": {
        "input_text": "Second task input",
        "category": "demo"
      }
    }
  ]
}

See the API Reference for full endpoint documentation, response format, and batch limits.