Introduction to AI Orchestration Workflows

Overview

AI Orchestration enables users to create automated workflows using configurable steps.

Each workflow consists of structured steps that execute actions such as calling APIs, running queries, invoking AI, integrating with external systems, or controlling execution flow.

Creating a Workflow

Step 1: Navigate to Workflows

Navigate to AI Orchestration → Workflows
Click Add Asset

Step 2: Define Workflow Details

Provide the following:

Workflow Name
Description (optional)

Once created, the user lands on the Designer tab.

Step 3: Define Workflow Steps

Once the workflow is created, the Designer tab contains a YAML editor for configuring and modifying the workflow definition. It also provides a Test Workflow button to validate and execute the workflow for testing purposes, and a Done button to save changes and exit the Designer.

Each step must include:

id
type
name
config
next (if applicable)

Workflow Inputs

Workflows can accept input parameters at runtime using the optional inputs section defined at the workflow level. This allows the same workflow to be reused with different input values when executed manually or through the API.

Workflow inputs can be supplied when executing a workflow manually or through the Execute via API capability.

Example:

inputs:
  - path: my_param_path
    param: my_param_name
    type: string
    required: false
    description: Just testing

Workflow inputs can be referenced within workflow steps using the payload object and the configured input path.

Example:

- id: start_step
  type: start
  name: Start Workflow

  config:
    message: "Input value is {{ payload.my_param_path }}"

Note: Workflow inputs are referenced using the input path through the payload object. For example, use {{ payload.my_param_path }} instead of {{ my_param_name }}.

To insert a step type:

Place the cursor under type: and enter /.
A dropdown list appears showing all available step types.

The step types are grouped into categories:

Process Steps
Prompt AI
External Systems
Flow Controls

Selecting a step from the list automatically inserts the corresponding structure into the workflow editor.

Step Types:

Workflow steps are grouped into the following categories.

Process Steps

Process steps perform core operations within the workflow.

Call API – Executes a REST API operation from a configured API Collection.
Execute Database Query – The Execute Database Query step allows you to run a SQL query against a configured data source.
Execute JDBC Operation – Executes a predefined JDBC operation from a configured collection.
Execute Code – Runs custom Python or JavaScript code within the workflow.
Send Email – Sends email notifications during workflow execution.
Monitor Status – Monitors external execution or system state and resumes the workflow based on defined conditions.

Prompt AI

Prompt AI steps allow workflows to interact with AI models.

DvSum AI Agent – Executes a configured DvSum AI agent and returns its response.
LLM Call – Sends a direct prompt to a large language model and retrieves the generated response.

External Systems

External system steps integrate workflows with third-party platforms.

Jira Cloud – Performs Jira actions such as creating or updating issues.
Microsoft Teams – Sends messages or notifications to Microsoft Teams channels.

Flow Controls

Flow control steps manage workflow execution logic.

Decision – Routes workflow execution based on conditions or value matching.
Fan-Out Workflow – Invokes another workflow as part of the current workflow.
Composite – Groups multiple steps into a single logical block.
For Each – Iterates through a list of items and processes them individually.
Human in the Loop – introduces a manual approval or intervention point within the workflow.

Workflow Menu Options

The three-dot menu provides additional management and execution options for a workflow.

The Export YAML option allows users to download the workflow configuration in YAML format.
The Version History option displays all saved versions of the workflow. Each version record shows details such as the version type (for example, draft or published), the user who saved the version, and the timestamp when it was saved. From the version history view, users can export the YAML of a specific version or delete that version if needed.

The Execute via API option provides the API endpoint that can be used to trigger the workflow programmatically. This allows systems to execute the workflow through an API call.

For more information on executing a workflow via API, please refer to the article Executing a Workflow via API

The Add to Job option allows the workflow to be added to a scheduled job. Once added, the workflow can be executed automatically based on the job’s configured schedule.

The Remove from Job option removes the workflow from an existing job. After removal, the workflow will no longer be triggered by that scheduled job.
The Clone option creates a duplicate copy of the workflow. This allows users to reuse an existing workflow configuration as a starting point for creating a new workflow without modifying the original.
The Delete option permanently removes the workflow from the system. Once deleted, the workflow configuration and its associated versions are no longer available.

In This Article: