Uploaded May 2026 | Updated September 2026, 2 weeks ago
Build Safe AI Agents in APEX: Learn How Function Calling Connects Models to Real Business Work
This session shows how AI function calling turns Oracle APEX from a chat interface into an application that can safely perform real work. Attendees will learn how to design AI tools and map function execution to APEX and ORDS components without exposing direct SQL access. The session also covers guardrails, validation, logging, and practical design patterns so you can build AI-assisted features that are useful, secure, and reliable.
You’ll learn from Jonathan Dixon, Oracle ACE Pro:
・How function calling works and why it matters more than prompt-only or retrieval-only AI patterns in APEX applications.
・How to design effective tools with clear schemas, narrow business actions, and a clean separation between read and write operations.
・How to implement the orchestration and dispatch pattern in APEX and PL/SQL to validate, authorize, execute, and return structured tool results.
・How to apply guardrails such as validation gates, row-level security, audit logging, and testing to make AI-powered actions production-ready.
Timestamps
0:00:00 - Welcome and Introduction & Announcements - Karina Pastor
0:04:10 - Jonathan Dixon Introduction
0:05:39 - Agenda Overview
0:06:44 - Understanding Agents and Their Components
0:09:00 - Importance of Function Calling
0:10:00 - Explanation of the Tool Calling Loop
0:12:18 - Tool/Function JSON Schema
0:15:29 - Parallel Tool Calling
0:17:21 - Demo of Order Support Copilot
0:22:48 - Analyzing Responses and Context in the Demo
0:25:49 - More Complex Demo Question
0:26:28 - Designing Effective Functions
0:29:04 - Latency and Performance Considerations
0:33:14 - Compaction of Data Sent to the Model
0:34:33 - Human Involvement in the Loop
0:36:22 - Keeping Agents Safe
0:41:16 - Security Best Practices
0:43:30 - Logging and Its Importance
0:44:53 - Second Demo: Project Management Application
0:50:28 - Key Takeaways and Closing Remarks
0:54:34 - Acknowledgments and Conclusion
-
Plus: GitHub Repo with the presentation and demo code:
github.com/cloudnueva/ai_function_calling_with_apex
Build Safe AI Agents in APEX: Learn How Function Calling Connects Models to Real Business Work
This session shows how AI function calling turns Oracle APEX from a chat interface into an application that can safely perform real work. Attendees will learn how to design AI tools and map function execution to APEX and ORDS components without exposing direct SQL access. The session also covers guardrails, validation, logging, and practical design patterns so you can build AI-assisted features that are useful, secure, and reliable.
You’ll learn from Jonathan Dixon, Oracle ACE Pro:
・How function calling works and why it matters more than prompt-only or retrieval-only AI patterns in APEX applications.
・How to design effective tools with clear schemas, narrow business actions, and a clean separation between read and write operations.
・How to implement the orchestration and dispatch pattern in APEX and PL/SQL to validate, authorize, execute, and return structured tool results.
・How to apply guardrails such as validation gates, row-level security, audit logging, and testing to make AI-powered actions production-ready.
Timestamps
0:00:00 - Welcome and Introduction & Announcements - Karina Pastor
0:04:10 - Jonathan Dixon Introduction
0:05:39 - Agenda Overview
0:06:44 - Understanding Agents and Their Components
0:09:00 - Importance of Function Calling
0:10:00 - Explanation of the Tool Calling Loop
0:12:18 - Tool/Function JSON Schema
0:15:29 - Parallel Tool Calling
0:17:21 - Demo of Order Support Copilot
0:22:48 - Analyzing Responses and Context in the Demo
0:25:49 - More Complex Demo Question
0:26:28 - Designing Effective Functions
0:29:04 - Latency and Performance Considerations
0:33:14 - Compaction of Data Sent to the Model
0:34:33 - Human Involvement in the Loop
0:36:22 - Keeping Agents Safe
0:41:16 - Security Best Practices
0:43:30 - Logging and Its Importance
0:44:53 - Second Demo: Project Management Application
0:50:28 - Key Takeaways and Closing Remarks
0:54:34 - Acknowledgments and Conclusion
-
Plus: GitHub Repo with the presentation and demo code:
github.com/cloudnueva/ai_function_calling_with_apex
