1. How do I use AI Agent Studio?
  2. Create and Enable Workflows for Using in Apps

Create and Enable Workflows for Using in Apps

An agentic app is built using agent teams of type Workflow. These agent teams provide the response when the agentic app requests for displays, actions, communications, summary text, subtitle text, or user-query answers. The recommended method is for the agent team to receive the request from the agentic app using the OraMessageHint webhook variable. Create your agent team structure to receive the webhook message, branch the workflow, and then provide the needed output to the agentic app.

To use workflows in agentic apps, you must first enable the workflow agent teams from AI Agent Studio.
  1. Go to AI Agent Studio and open the Agent Teams tab.
  2. Edit the workflow agent team you want to use in agentic apps.
  3. Select Agent Team Settings Agent Team Settings Icon.
  4. From the Details tab, select Expose to agentic apps.
  5. Save your changes.

OraMessageHint Variable

OraMessageHint is the primary app-aware workflow variable, and its value can be retrieved using the expression $context.$app.$OraMessageHint. It tells your workflow why the app is calling it. In prompt templates, use expressions such as {{ $context.$app.$OraMessageHint }}. In Code nodes, use the value as $context.$app.$OraMessageHint.

Here are some common values for OraMessageHint:

Value Description
Summary Indicates that the app is asking for summary text. Return short summary text for that agent's contribution.
InitSubtitle Indicates that the app is asking for the dynamic subtitle. Return the subtitle text itself.
InitDisplay Indicates that the app is asking for startup displays. Return one or more oraInfoDisplay blocks, and it may also include actions and communication suggestions when that improves the startup experience.
InitActions Indicates that the app is asking for startup actions. Return zero or more action suggestions.
InitCommunications Indicates that the app is asking for startup communication suggestions. Return zero or more oraComms blocks.
Query Indicates that the user asked something through the app. Use the input message and answer the user. This response can also include displays, actions, and communications when appropriate.
InvokeAction Indicates that a UI action sent work back to the workflow. Use this for follow-up handling after a widget command triggers an action path that sends a command to an agent.
AdditionalContent Indicates that the app is loading an extra panel under a focused agent. Use OraPanelName to tell which panel is being requested.
FillParameters Indicates that the app is asking one or more agents to fill communication parameters before generating or sending a response.
SendCommunication Indicates that the app is asking a target workflow to send or complete a communication flow.