Your first workspace

Learn the building blocks and plan a small, testable pipeline.

StreamSource guidesUpdated 12 October 2026

Before you start

Choose one source, one event type and one destination. A small example makes it easier to tell whether each part of the pipeline is doing its job.

  • A sample event with no credentials or personal data.
  • A source you can inspect, such as an MQTT topic or a database change stream.
  • A destination that you can query or inspect.
  • A clear result: what should appear at the destination?
The current website is a frontend preview. Account creation and deployment are not connected to a backend. This guide describes the workflow to plan and validate, rather than commands you can run against a live StreamSource service.

Understand the building blocks

ConceptPurposeExample
WorkspaceA boundary for a team’s pipelines and access.fleet-demo
SourceReceives events from an external system.MQTT vehicle telemetry
TopicThe stream between components.fleet.raw.telemetry
TransformChanges or enriches each event.Normalize speed and timestamp
SinkDelivers processed events.PostgreSQL or a webhook

Start with a separate development workspace and sample credentials. Keep production inputs out of your first experiment.

Plan your first workspace

  1. Name the workspace.

    Pick a stable name that describes the team or use case, such as fleet-demo.

  2. Define the event contract.

    List required fields, their types and the identifier you will use to track an event.

  3. Map the pipeline.

    Draw source → topic → transform → topic → sink. Decide which component owns each conversion.

  4. Choose a success check.

    Define the expected destination record before connecting a real source.

Use the connector catalogue to check available, beta and planned integrations.

Prepare a sample event

This fictional event represents one vehicle reading. The same structure is used in the pipeline guide.

JSON example
{
  "event_id": "evt-demo-001",
  "device_id": "vehicle-101",
  "speed": "42.5",
  "recorded_at": "2026-10-12T08:00:00Z"
}

The speed arrives as a string. The transform will convert it to a number and rename fields consistently.

Verify before expanding

  • The source receives the sample event.
  • The transform produces the expected output fields.
  • The destination contains the expected record.
  • A retry does not silently create an unwanted duplicate.
  • An invalid event is visible and can be investigated.

Once you can explain the path of one event, add more sources, destinations or business logic.

Continue learning