Implementing Agentic Apps

Use this topic to understand the implementation pattern for agentic apps in Oracle Permitting and Licensing. Agentic apps provide a task-focused workspace that brings together AI-generated analysis, recommendations, supporting information, and user-initiated actions. Instead of using a single AI Assist button or opening separate agents, users work in one workspace where multiple workflow agents can help summarize information, recommend next steps, prepare communications, and support actions.

For setup details for a specific app, use the setup topic for that app. The setup topic identifies any required Business Rules Framework rule, workflow configuration, schedule, document, or workspace configuration.

Oracle Permitting and Licensing delivered the Code Enforcement Incident Workspace, an agentic app that helps code enforcement teams review, prioritize, and act on incoming incidents from a single workspace. The workspace brings together incident analysis, recommended actions, active incident summaries, and communication support. For more information, see

Working with Agentic Apps

An agentic app coordinates one or more workflow agents to support a business process. The app presents output from those agents in a workspace and lets users take action from the same experience.

An agentic app can include:

Summaries or insights generated by workflow agents.
Recommendations based on transaction data and configured workflow logic.
Action buttons that let users initiate a business process.
Communication support, such as generated email text that users review before sending.
Panels, charts, or lists that help users monitor workload or prioritize work.

The app can recommend and prepare an action, but the user initiates the action from the workspace. In the Code Enforcement Incident Workspace, for example, the app can recommend actions such as closing a duplicate, updating priority, or scheduling an inspection, but the user selects the action before the workflow agent executes the related business process.

Understanding Agentic App Implementation

Implementation varies by app, but most agentic apps follow this pattern:

Enable AI functionality for the applicable offering.
Configure AI security for administrators and users.
Confirm that the app’s supporting workflow agents are delivered or configured.
Confirm that any required workflow agents are published and available for the app.
Configure any app-specific trigger, rule, schedule, or workspace setup described in the app setup topic.
Verify the workspace with a user who has the correct product and AI access.

When the agentic app as delivered enabled, setup may only include enabling AI for the offering and configuring security. Agentic apps may have different setup requirements.

Understanding Workflow Agents Used by Agentic Apps

Agentic apps rely on workflow agents to perform analysis, prepare content, and execute user-initiated actions. Depending on the app, workflow agents can:

Analyze a record or transaction.
Generate a workspace summary.
Prepare panel content or charts.
Generate communication text.
Execute an action selected by the user.
Store AI-generated results for later display.

For example, the Code Enforcement Incident Workspace currently uses several workflow agents: Code Enforcement Incoming Incident Evaluation Assistant, Code Enforcement Active Incident Summary Generator, and Inspection Scheduling Agent. These agents support reported incident recommendations, active incident summaries, priority actions, communication content, and inspection scheduling.

For a specific agentic app, use the app setup topic to confirm which workflow agents are required and whether any app-specific workflow configuration is needed.

Using Stored Analysis and Business Rules Framework

Some agentic apps may depend on AI analysis that runs before the user opens the workspace. For example, the Code Enforcement Incident Workspace displays analysis that is performed when an incident is submitted and stored for later use in the workspace.

When an app uses stored analysis, insight into the implementation might include:

What event or process creates the AI analysis.
Whether a Business Rules Framework rule invokes the workflow agent.
Where the analysis appears in the workspace.
Whether refreshing the workspace reruns the analysis or only reloads stored results.
How to regenerate the analysis, if needed.

For more information, see Business Rules Framework setup topics when a rule is required.

Securing Agentic Apps

Users need AI runtime access and the standard product access required for the business process. They also need permission to perform any actions exposed in the workspace.

For example, a user who can view an incident recommendation might still need the appropriate Code Enforcement access to update priority, close an incident as a duplicate, schedule an inspection, or send a communication.

For more information, see Securing AI Agents and Agentic Apps. See the setup topic for the specific app to identify product-specific roles or duty roles.