Join us for a deep dive into the evolution of artificial intelligence, moving beyond passive interactions to building autonomous systems. Titled "Let Him Cook," this session focuses on "Architecting AI with Gemini" and exploring the paradigm shift from "Passive Models to Agentic Workflows".

The Menu

The session is structured into four key segments designed to take you from foundational concepts to practical application:

• The Ingredients: A breakdown of the core technologies, including AI, Machine Learning, Deep Learning, and Generative AI.

• The Shift: Understanding the transition from standard Large Language Models (LLMs) to Agentic AI.

• The Kitchen: A look at the tools required to build agents, including Retrieval-Augmented Generation (RAG) and external tooling.

• The Build: A live demonstration of constructing an agentic workflow.

Key Concepts

Attendees will learn to distinguish between standard LLMs and Agentic AI. While standard LLMs function as a "Brain in a Jar"—holding vast, frozen knowledge but remaining isolated—Agentic AI acts as a "Brain with Hands". These agents wrap the LLM with the ability to use tools, browse the web, execute code, and access databases to achieve specific goals rather than just predicting text.

The Architecture of Gemini

The session will explore how Gemini functions as "The Oven" for these agents, utilizing:

• Massive Context Window: Leveraging Gemini's 1M+ token capacity to process entire codebases or books and maintain state for complex tasks.

• Native Multimodality: The ability to process and act on visual information (video and images) alongside text.

• Memory & Grounding: Using Vector DBs and embeddings for memory (RAG) and connecting models to Google Search to verify facts and eliminate hallucinations.

• Action: Implementing function calling to allow agents to execute tasks, such as writing files to a system.

Prerequisites:

• Node js version v20 & above and npm v10 & above.

• Familiarity with JavaScript/Node.js is recommended but not required.