Published at: 2026-09-17

Create a flow

Use Flow Designer to orchestrate complex automation logic.

Overview

A Flow is suitable for complex automation that requires variables, Branches, loops, interface interactions, data reads and writes, or AI Nodes. Admins can drag Nodes in Flow Designer, connect execution paths, and pass context through variables.

Before you begin

  • You have Flow management permission.
  • You have Confirmed that the Business Scenario is suitable for a Flow.
  • You have Confirmed whether a user interface is required.
  • You have prepared test data.
  • If you need to call functions, models, or Knowledge Management, Configure the corresponding resources first.

Step 1: Go to Flow management

  1. On Setup > Process Management > Flow Management, click New.
  2. On the Basic Info page, enter the Process Name.
  3. In Process Type, select the Flow Type.
  4. Enter the Process Description.
  5. Click Next.

Step 2: Select the Flow Type

Flow Type English Name Use case
Screen Flow Screen Flow A user clicks a button, and the Flow must show an interactive page and collect input
Auto Flow Auto Flow No user interface is required. The Flow automatically executes a set of tasks
Agent Action Flow Action Flow Custom Agent actions in AI agent scenarios
Agent Flow Agent Flow Orchestrate an Agent through Flow
Prompt Template Flow Prompt Template Flow Use Flow to help generate prompt templates

Step 3: Understand Flow Designer

Flow Designer has five areas:
Area Purpose
Toolbar area Undo, redo, multi-select, move canvas, align, zoom, reset, Delete, and Export
Process Node area Drag start, end, Branch, loop, assignment, function, data operation, and other Nodes
Resource management area Create, Edit, and Delete variables, and View variable references
Canvas area Configure Nodes and Node connections
Node configuration area Configure Node Name, Description, and dedicated parameters
[!NOTE] Interface interaction Nodes can be used only in Screen Flows. They are not Available in Auto Flows.

Step 4: Configure variables

  1. In Resource management area, click New variable.
  2. Enter the variable Name.
  3. Select the variable Type.
  4. Set a default value as needed.
  5. Click Save.
Variables Support basic Types and composite Types.
Category Type Description
Basic Type Text Stores names, descriptions, message Content, and other Text
Basic Type Numerical Value Stores integers or decimals
Basic Type Boolean Stores yes or no
Basic Type Array Stores multiple basic Type elements
Basic Type Dictionary Stores Key-Value data
Composite Type Object Stores business Object records. Single and Multi-Select values are Supported
Composite Type Personnel Stores Personnel IDs. Single and Multi-Select values are Supported
Composite Type Department Stores Department IDs. Single and Multi-Select values are Supported
Composite Type File Stores File information. Single and Multi-Select values are Supported
Composite Type Date Stores Date
Composite Type Date Time Stores Date and Time
[!WARNING] After a variable is referenced by a Node, the System does not allow direct Delete. Clear references before Delete.

Step 5: Configure the start Node

  1. In Canvas area, click the Start Node.
  2. Enter the business Name.
  3. Enter the Business Description.
  4. In Input, Add Process input parameters.
  5. Click Save.
The start Node defines inputs for the whole Flow. Input Fields Support basic Type variables and composite Type variables.

Step 6: Add Process Nodes

  1. In Process Node area, select a Node.
  2. Drag the Node to the canvas.
  3. Click the Node to open the right-side configuration area.
  4. Configure the business Name and Business Description.
  5. Configure Node-specific parameters.
  6. Connect Nodes on the canvas.
  7. Click Save.

Common Node configuration reference

End Node

An End Node outputs the final result of a Flow. Every complete execution path should connect to an End Node. If the Flow has multiple conditional Branches, you can Configure multiple End Nodes.
Configuration items include business Name, Business Description, and Node output. Output Fields can reference output variables from any Upstream Node.

Parallel Node

A Parallel Node executes multiple Branch tasks at the same Time. It requires one input path, Supports multiple output paths, and does not require extra parameters.

Branch condition Node

A Branch condition Node enters different paths based on conditions. Each output path can Configure conditions.
Supported conditions include:
  • Include.
  • Not Included.
  • Is Null.
  • Is Not Null.
  • Equals.
  • Not Equal To.
When no Branch condition is met, Configure a default path. Otherwise, the Flow may report an error at runtime.

Loop Node

A Loop Node iterates over a collection variable. Admins must select a collection variable as input and Configure loop order.
Loop order Supports:
  • Loop from the first data item.
  • Loop from the last data item.

Assignment Node

An Assignment Node assigns values to variables. Each assignment group handles one variable. Admins can Add multiple assignment groups and use fixed values or variables of the same Type as value sources.

Interface interaction Node

An interface interaction Node displays a page and collects user input.
Required configuration includes:
  • Business Name.
  • Select Object.
  • Operation Type.
  • Select variable.
  • Output.
Operation Type Supports Edit and New. The selected variable Type must Match the selected Object Type.

Function Node

A Function Node executes Groovy, Java, or APL extension logic. It is suitable for structured data processing, mathematical calculation, and data concatenation.
During configuration, click Add APL code, select an existing function or create a function, and Configure input and output variables.

Template conversion Node

A template conversion Node processes Text, JSON, or XML fragments. It Supports jinja2 template syntax and can Insert all variable Types.

Large model Node

A large model Node calls LLM capabilities. It can be used for intent recognition, Text generation, Content classification, Text translation, code generation, RAG, Image understanding, and File analysis.
During configuration, select a prompt template and bind Flow variables to prompt input and output parameters.

Business operation Node

A business operation Node executes business capabilities, including sending CRM notifications, Knowledge Management Q&A, semantic retrieval, integration Flows, and converting document Files to Markdown.
When sending CRM notifications, required parameters include business Name, Reminder Personnel, Reminder Title, and Reminder Content.

Create data Node

A create data Node creates Object data based on Flow variables.
Creation methods include:
  • Create by specified Fields.
  • Create by specified Object variable.

Update data Node

An Update data Node Updates Object data based on Flow variables.
Update methods include:
  • Update by specified Fields.
  • Update by specified Object variable.
  • Update by specified Object collection variable.

Get data Node

A get data Node queries Object data and stores results in variables. Admins can Configure query Object, query conditions, Sorting rules, and query Fields.

Expected result and validation

  1. Save the Flow in Flow Designer.
  2. Click Debug.
  3. Enter test parameters.
  4. View the execution result of each Node.
  5. Confirm that End Node output meets expectations.

Related topics

Submit Feedback
Next
Publish and debug flows