Designing Intelligent AI Agents Mastering
System Prompts and Topics
in Oracle AI Studio
As organizations strive to automate complex business processes, Oracle AI Studio offers a powerful platform for creating AI-driven agents that are smart, reliable, and aligned with enterprise needs. At the core of every successful agent lies a thoughtful configuration of System Prompts and Topics—the foundational elements that define how agents behave, what they can access, and how they respond.
This article explores the essential components of working with System Prompts and Topics, with real-world context from the “LD EMP AGENT” configuration. Understanding these components will enable business and technical teams to build highly focused and capable AI Agents that deliver consistent value.
What Is a System Prompt?
A System Prompt serves as the brain of an AI Agent. It determines:
- The Agent’s persona—how it should speak, think, and interact.
- The capabilities and tools it can access.
- The boundaries of its knowledge and functionality.
- The tone, logic, and structure of the responses it produces.
Each Agent in Oracle AI Studio is defined by a system prompt that provides clear guidance on how it should handle tasks and what to avoid. This is not a simple script—it’s a strategic instruction set that enables the Agent to reason effectively.
Example:
In the “LD EMP AGENT” (Worker Agent) example:
- The prompt defines the Agent as a helpful assistant responsible for answering questions about a worker’s personal and employment details.
- It explicitly instructs the Agent to avoid generating facts and to rely only on accurate, verified tool call responses.
- Tools like Add Assignment, Change Manager, Promote and Change Position, and others are referenced, showing what actions the Agent is allowed to assist with.
Topics: Defining the Scope, Intent, and Purpose
Topics further refine the system prompt by specifying the scope, intent, and purpose of the Agent. While the system prompt shapes the Agent’s overall behavior, topics direct the Agent toward certain business domains or tasks.
This ensures that the AI Agent does not attempt to handle tasks beyond its intended design. In Oracle AI Studio:
This ensures that the AI Agent does not attempt to handle tasks beyond its intended design. In Oracle AI Studio:
- Topics are added to the system prompt to provide focus.
- They serve as filters, telling the LLM what kind of questions are within scope.
- They help route responses, especially in multi-agent or agent team configurations. In the LD EMP AGENT example:
- The topic focuses solely on worker employment data.
- If a user asks a question outside this topic, the Agent is instructed to politely decline the query and clarify the topics it can handle.
Templates: Reusability with Flexibility
Oracle provides pre-built templates that include a combination of System Prompts and Topics. These templates are crafted for common enterprise use cases and can be:
- Used as-is for rapid deployment.
- Customized to fit specific operational needs.
- Extended with new prompts or topics to match evolving business logic.
These templates allow faster configuration and lower the barrier to creating effective AI Agents. Teams can build upon best practices embedded in Oracle’s templates while tailoring the behavior to suit their business requirements.
Special Instructions:
- For out-of-scope queries, the Agent must politely decline and explain what it can handle.
- JSON responses must be well-formed and valid, with proper curly braces.
- The Agent must not make assumptions or fabricate data.
- When generating multiple responses, tools must be called sequentially—no nested tool calls.
- In multi-person queries, the Agent is instructed to answer individually for each person using available Person ID and Assignment ID details.
System Prompts for Different LLMs
Different Large Language Models (LLMs) interpret prompts and instructions differently. Oracle AI Studio allows users to optimize prompts for specific LLMs by adjusting:
- The format and structure of the prompt.
- The instruction style or tone.
- The level of detail required for accurate processing.
This is critical for maximizing performance, especially when deploying Agents across multiple models or adapting to updated versions of existing LLMs.
Real-World Example: LD EMP AGENT
The LD EMP AGENT configuration shown in Oracle AI Studio demonstrates the practical application of System Prompts and Topics:
Agent Purpose:
To assist with queries related to a worker’s personal and employment details.
Authorized Actions:
The Agent is allowed to guide users through tasks using specific tools, such as:
- Add Assignment – Add an assignment to a worker's record.
- Change Assignment – Change a worker’s assignment.
- Change Location – Update a worker’s job location.
- Change Manager – Assign a new manager.
- Change Work Schedule – Modify a worker’s hours.
- Global Transfer – Manage local/global transfers.
- Manage User Account – Make changes to user account access.
- Promote and Change Position – Promote a worker and change job roles.
- Terminate Employment – Terminate a worker’s employment.
- Transfer – Move a worker to another department.
Conclusion
Building effective AI Agents is not just about deploying a chatbot—it’s about crafting a thoughtful, purpose-driven experience. With System Prompts and Topics, Oracle AI Studio provides a precise and flexible framework for doing just that.
Whether you’re deploying an agent for HR, finance, or supply chain operations, using structured prompts and clearly scoped topics ensures that your AI Agent:
- Delivers accurate answers
- Maintains enterprise boundaries
- Aligns with business goals
- Performs consistently across tasks and users
Oracle AI Studio empowers you to create agents that think clearly, act purposefully, and add measurable value to your organization.
