Create Workflows

A workflow consists of a sequence of nodes used for deterministic, rule-based orchestration of tasks in which every step is preconfigured. Workflows are ideal for scenarios where compliance, repeatability, and governance are essential. The workflow tasks are represented as a connected sequence of nodes. Each node within the workflow performs a defined function, such as extracting data, calling a business object function, running an LLM, or sending an email. The output from each node is seamlessly passed to the following step, ensuring a controlled and efficient process flow.

When designing your workflow, keep these process breakdown steps in mind:

Model the manual workflow. Start from a high-level process and break it down into smaller, automatable steps.
Identify task dependencies and group nodes by type. For example, distinguish between back-end logic tasks and those driven by LLMs. Doing so helps to streamline operation and minimize unnecessary LLM usage.
For each task, define the node type, required input values, expected outputs, and any transformation logic needed.
For recommendations about improving your agent's efficiency, see How do I make agents respond faster?

Let’s look at an example to illustrate how a workflow operates in practice. The following workflow is designed to help buyers manage order inquiries within predefined constraints. It allows buyers to retrieve details about their existing orders, including order status, shipment tracking, payment status, and order history. The workflow provides real-time updates on order progress and answers specific questions regarding terms, delivery dates, and other order-related inquiries. For complex issues or inquiries that fall outside its predefined capabilities, the workflow seamlessly escalates the request to a sales representative, ensuring that the buyer receives the necessary support while maintaining an efficient self-service experience.

Example Workflow

Workflow Example

Step	Node Name	Description
UserIntent	LLM	Classifies the request as create, enquire, modify, cancel, or not supported.
UnsupportedReturn	Return	Returns the unsupported request response.
UnsupportedIntentOutput	LLM	Generates a buyer-friendly response for unsupported requests.
Start	Start	Begins the workflow.
setBuyerContext	Set Fields	Loads the buyer profile, default order attributes, and supported intents into workflow variables.
RetrieveOrderStatusTypes	Document Retrieval	Retrieves valid sales-order status values from the reference document.
Response	LLM	Wraps the buyer-facing content into the standard response JSON.
requiresLineItemDetails	Condition	Determines whether detailed line-item information is required.
QueryGenerator	LLM	Converts the buyer inquiry into a structured output and identifies support or additional information needs.
ProcessInput	Code	Extracts the current user message from the incoming interaction.
OrderIntentRouter	Switch	Routes the create-order subflow.
isValidRequest	Condition	Checks whether the classified intent is supported.
IsSupportedQuery	Condition	Checks whether the enquiry is supported.
IntentRouter	Switch	Routes supported requests to the Create, Enquiry, Modify, or Cancel flows.
getOrderableItems	Business Object	Searches the orderable product catalog.
Get Sales Order Details	Business Object	Retrieves sales-order headers, totals, and line summaries.
Fetch Sales Order Lines Details	Business Object	Retrieves detailed line information for the first matching sales order.
ExtractOrderInfo	LLM	Extracts product, quantity, order-line information and buyer defaults; determines whether to create an order, perform product discovery, or request missing fields.
ExtractModifyOrderInfo	LLM	Extracts order header ID, product ID, and requested quantity.
ExtractModifyIntent	LLM	Determines whether the modification is to add or remove a line item.
ExtractCancelOrdreNumber	LLM	Extracts the sales order number to cancel.
End	End	Completes the workflow.
DSL Query Generation	LLM	Generates Elasticsearch query clauses for product discovery.
CreateOutput	LLM	Converts processing results into buyer-friendly content.
createOrderFunction	Business Object	Creates a sales order using buyer defaults and extracted order details.
createErrorOutput	LLM	Explains missing mandatory order fields.
ClarificationOutput	LLM	Generates a clarification message describing supported enquiry options.
Cancel Order	Business Object	Cancels the specified sales order.
AskClarification	Return	Returns the clarification response.
addNewLineToOrder	Business Object	Adds a new product line to an existing sales order.
Add (False Branch)	Pass-through	Passes control to the next node on the false branch.
Add (False Branch)	Pass-through	Passes control to get sales order details.
Add (False Branch)	Pass-through	Passes control to create output.

Here are the steps to create your own workflow.

Go to AI Agent Studio.
From the Workflows tab, select Add.

Configure the workflow.

Details Tab

Field	Description
Agent Team Name and Code	Enter Agent Team Name and Code. Use letters, numbers, and underscores for the code.
Family	Select the family to which this agent team belongs.
Product	Select the product within the family to which this agent team belongs.
Description	Add a description of the workflow.

LLM Tab

Field	Description
Provider	Choose to use the default model or select a model. When using a custom model, specify the model properties.

Security Tab

Field	Description
Add	Select the roles which will have access to this workflow.

Chat Experience Tab

You can enable users to upload files directly in chat or from third-party providers and receive responses in their preferred language.

These are the supported file types users can upload to the chat: PDF-Tagged / Scanned, HTML, JSON, MARKDOWN, XML, TXT, DOCX, DOCX-Multimodal, XLSX, CSV, PPTX, JPG, PNG, and ZIP.

Field	Description
Answer in user's language	Select this option to allow users to receive answers in their preferred language.
Enable file upload	Select this option to allow users to upload files from local storage.
Enable third-party file upload	Select this option to allow users to upload files from connected cloud storage accounts. To enable this functionality, add credentials for your provider by navigating to the Credentials tab in AI Agent Studio and completing the Chat Experience setup.

Questions Tab

Field	Description
Starter Questions	Enter initial questions for the workflow.
Follow-up Questions	Enable this option to indicate that the workflow can ask follow-up questions based on the user' s conversation history.

Triggers Tab

Create triggers to begin the workflow. Triggers define the conditions or events that start the workflow and identify when and why it begins.

Field	Description
Type	Add the data type of the trigger.

For more information about setting up triggers, see Configure Workflow Triggers.

Variables Tab

Define workflow-level variables, making them accessible to all nodes within the workflow. Variables are ideal for storing IDs, constants, or any values you need to share and use across different parts of the workflow.

Field	Description
Type	Add the data type of the variable.

Error Handling Tab

Specify where to send an email when a workflow instance runs into a permanent error. You can also use context expressions in these fields for more clarity.

Add nodes to the workflow.
Tips
- Connect nodes in sequence to control how input values are received, processed, and forwarded throughout the workflow.
- You can insert new nodes before, after, or between existing nodes as needed.
- For more information about Vector nodes, see Configure Vector DB Writer and Reader Nodes in a Workflow.
Publish the workflow.

Users can view the published workflows from the Agents and Applications page. To access this page, go to Me > Quick Actions > Show More > AI Agent Studio > AI Agent.
Note: Make sure that your users have access to interact with the AI agents. For information, see How can I give users access to AI Agents?