Скачать или смотреть KEYNOTE PRESENTATION: Smarter AI, Better Decisions - Planning Engines Meet Generative Models

As enterprises push the boundaries of generative AI, one key hurdle remains: how to coordinate multiple AI tools and models to solve real-world business problems, intelligently and at scale.

This session will showcase how UC San Diego’s agentic AI architecture tackles that challenge by extending the Model Context Protocol (MCP) with a task planning engine. The result? A more powerful, modular system that integrates local and cloud-based LLMs with planning, RAG, and analytics capabilities.

Learn how this enhanced framework empowers an Enterprise Data Agent (EDA) to dynamically select and execute the right tools (using metadata-driven logic) to deliver real-time answers and insights. Examples will include dynamic visualizations, text-to-SQL conversion, and portfolio risk analytics powered by SAP HANA S/4 Activity Hubs.

You will also learn:

Why embedding planning logic in generative systems drives smarter task execution
How MCP extensions enable scalable orchestration of AI tools across domains
Real-world examples of AI-enabled fund and portfolio management in action