Skip to main content
In MantrixFlow, a pipeline branch is a path on the builder: source → (optional) transform → destination. You can add more branches to send the same (or filtered) data to another destination or to apply different transforms. This is not a separate “environment” or “Git branch” screen—branching is modeled on the canvas only.

​
Open the transform editor

  1. Open Data Pipelines → select your pipeline → Open Builder (or click the pipeline row).
  2. Click the transform node (e.g. labelled with PYTHON and a short code preview).
  3. Use Transform settings (gear) on the node, or click the node body to open the side panel.

​
Configure branch label, table, script, and errors

The transform panel includes:
  • Branch label — shown on the canvas; rename to match the destination or use case (e.g. “Analytics warehouse”, “Support replica”).
  • Source table — which stream/table this transform receives.
  • Python script — must define transform(record) (or the pattern shown in the template). Return a dict matching the shape you want downstream.
  • On error — Stop run on error vs Skip record and continue.
  • Preview with sample data — validates logic against sample rows.
  • Save Script — persists changes to the pipeline graph.
Transform panel with Python script and branch label

​
Add another branch (destination path)

On the canvas:
  • Use Add transform after the source (or between nodes) when you need a new transform step.
  • Use Add emitter (destination) / Add destination to attach another destination connection for a parallel branch.
Each branch can target a different destination connection, schema/table, and append / merge / overwrite (emit) behaviour configured on the destination node (Destination / emit settings). Builder showing source, transform, and destination branch

​
AI-assisted edits

Ask AI on the builder toolbar can suggest graph changes; always review suggestions before saving or running in production.

​
Operational tips

  • After editing transforms, Save on the builder and run a test sync or small manual run before scheduling.
  • If a run fails, use Run History on the pipeline detail page and the Activity Log for correlated events.