How to create your first Weave flow?

A Weave is a visual pipeline that pulls data from one or more sources, transforms it through a chain of nodes, and writes the result to a destination. This article walks through building your first Weave end to end: adding a source, transforming, inspecting the result, saving, and running it.

Note: “Weave” is the current name for this pipeline. Earlier interfaces called the same thing a “Workflow.” They are the same feature; this article uses “Weave” throughout.

Before you begin

You need the Weave Designer role. Have at least one source ready to import from: a configured Datasource, a Connector, a saved Custom Script from Lens, or another Weave’s Result output. A Weave flows from source nodes (Data Import) through transformation nodes (such as Cleansing, Imputation, Duplicates) to terminal nodes (Result for inspection, Output for persistent writes).

Two habits make authoring smoother. First, each node’s configuration panel opens on the right and follows a Save / Discard pattern. Commit with Save or revert with Discard; closing without choosing triggers a warning. Second, you can run a single node while authoring to test it in isolation, separately from running the whole canvas.

Steps

Open the Weave canvas

1. Open Design → Add New → Weave. A new canvas opens with the node panel on the top and the configuration panel on the right.

Add a source node (Data Import)

1. From the node panel’s Data category, drag a Data Import node onto the canvas.
2. Select the node to open its configuration on the right, then choose a Data Source Type: Datasource, Custom Script, Connector, or Data Weave. The Data Source Type reference is on the companion Product Documentation page.
3. Configure the source.
4. Select Save in the configuration panel.

Add and connect a transformation node

1. From the Data Preparation category, drag a transformation node (for example Cleansing, Imputation, or Duplicates) onto the canvas.
2. Connect the nodes: click the tip of the Data Import node and drag the line to the transformation node.
3. Select the transformation node, configure it on the right, and select Save.

Inspect the data (Result)

1. To test as you go, select a node and select its Run icon to run that node in isolation.
2. Alternatively, select the canvas-level Run button in the toolbar to run the connected nodes in dependency order (Design Mode). Output appears in the Result Tab at the bottom.
3. In the Result Tab, use the column-header filter icon to narrow values, and drag a column divider to resize.

Add a terminal node

1. For design-time inspection only, add a Result node and connect it downstream.
2. To write the data to a destination, add an Output node, connect it downstream, configure the destination, and select Save.
3. A Weave can have multiple terminal nodes and several Result nodes at different branch points, or several Output nodes to write to multiple destinations.

Name, save, and run the Weave

1. To rename any node, select it, select the Pencil icon, and enter the new name. To remove a node, select it and select the Cross icon (reconnect any downstream links afterward).
2. Save the Weave from the toolbar Save action and give it a name. The saved Weave appears under Libraries.
3. To run in production, open Libraries → Weave, find your Weave, and select Run under Actions. This runs on cluster (Databricks) compute, as opposed to the design-time run on the canvas.

Result

You have a saved Weave that imports from your source, transforms the data, and (if you added an Output node) writes it to a destination. It appears in Libraries, where it can be run on cluster, scheduled, shared, or deleted. Design-time runs show output in the Result Tab; cluster runs execute the pipeline in production.