Skip to Content

Parallel & Merge nodes

The Flow Control category includes two nodes for concurrent execution:

NodeDescription
ParallelFork execution into multiple concurrent branches
MergeJoin parallel branches back together

Parallel node

The Parallel node splits execution into multiple branches that run concurrently.

Configuration

Click the Parallel node to open its configuration panel.

Parallel node configuration showing merge strategy, branches list, and output variable
Parallel node configuration showing merge strategy, branches list, and output variable
SettingTypeDescription
Node LabelText field (default: Parallel)Display name on the canvas
Merge StrategyToggle buttonsHow to handle branch completion (see below)
BranchesList + “Add Branch” buttonDefine named branches with optional conditions
Output VariableText field (default: parallel_results)Variable holding combined branch results

Merge Strategy options

StrategyDescription
Wait AllWait for all branches to complete before continuing
Wait AnyContinue as soon as any single branch completes
Wait NContinue after N branches complete
First SuccessContinue with the first branch that succeeds (others are cancelled)

Branch configuration

Each branch has:

  • Name — Text field for the branch label
  • Disable — Button to temporarily disable a branch
  • Remove — Button to delete the branch
  • Condition — Optional condition that must be true for this branch to execute

Advanced Options

Expandable section with timeout and error handling settings.

Edge significance

The Parallel node has N output edges — one for each branch. Each edge is labeled with the branch name.

Merge node

The Merge node joins parallel branches back into a single execution path. It’s the counterpart to the Parallel node.

Configuration

Click the Merge node to open its configuration panel.

Merge node configuration showing expected branches, aggregation mode, filter, and output variable
Merge node configuration showing expected branches, aggregation mode, filter, and output variable
SettingTypeDescription
Node LabelText field (default: Merge)Display name on the canvas
Expected BranchesNumber input (default: 2)Number of incoming parallel branches to wait for
Aggregation ModeToggle buttons (6 options)How to combine branch results (see below)
Filter ResultsCombobox (default: All Results)Filter which results to include
Output VariableText field (default: merged_results)Variable name for merged output

Aggregation Mode options

ModeDescription
ArrayCollect results into an array
ObjectCombine into a keyed object
FirstUse the first completed result
LastUse the last completed result
MergeDeep/shallow merge all results
CustomCustom aggregation expression

Edge significance

The Merge node accepts multiple input edges (one from each parallel branch) and has a single output edge continuing the workflow.

ℹ️

The Merge node is optional when using the Parallel node with a built-in merge strategy. Use it when you need explicit control over how branches rejoin or when branches were created manually without the Parallel node.

Common patterns

  • Fan-out/Fan-in — Parallel → multiple processing branches → Merge → continue with combined results.
  • Race condition — Parallel (First Success) with multiple API calls to different providers, taking the fastest response.
  • Partial results — Parallel (Wait N) when you only need responses from a subset of branches.