Data Streams
Transfer large volumes of records from one system to another with a source, an optional field-mapping processor, and a destination.
Coming soon
A Data Stream is a workflow step built for bulk record transfer — syncing contacts between CRMs, exporting rows to a warehouse, migrating data between systems. Instead of looping over items one action at a time, the Data Stream node reads records from a source, optionally transforms each record through a processor, and writes them to a destination as a managed batch job.
Like the AI Agent, the Data Stream node is a cluster root: its source, processor, and destination are cluster elements hosted inside the node rather than separate workflow steps.
The slots
| Slot | Role |
|---|---|
| Source | Where records are read from — a component operation that streams items (e.g. read rows, list records). |
| Processor (optional) | Transforms each record in flight. The standard processor is the Field Mapper, which maps source fields onto the destination's shape. |
| Destination | Where records are written — a component operation that accepts items in bulk. |
Simple vs Advanced editor
Clicking the Data Stream node opens its editor, which has two modes with a toggle in the header:
-
Simple editor (default) — a step-by-step wizard whose top nav walks through four steps in order: Source (pick the source component and operation), Destination (pick the destination component and operation), Mapping (the Field Mapper's source-to-destination field assignments), and Test (run the stream). Each step unlocks once the previous one is configured. On the Mapping step, Auto-map matching fields fills in assignments where source and destination field names line up. It's all done without touching the underlying element graph.
-
Advanced editor — the raw cluster-elements canvas, where source, processor, and destination appear as wired nodes. Use it to configure element parameters directly or to swap the processor for something other than the Field Mapper.
The simple wizard is available only while the processor slot holds the standard Field Mapper. If you replace it with a different processor in the advanced editor, the node opens in advanced mode from then on — the wizard has no way to represent a custom processor.
Both modes edit the same configuration, and your last mode choice is remembered per node.
How is this guide?
Last updated on