2025 ODSC West Workshop: Rethinking RAG: Building Smarter AI Agents with Agent2Agent and MCP

Welcome to the landing page for the workshop Rethinking RAG: Building Smarter AI Agents with Agent2Agent and MCP at the `2025 Open Data Science Conference (ODSC) West.

What to Expect

This workshop intends to provide an introduction to:

• Introduction to using Agent2Agent (A2A) Protocol to connect multiple agents together
• Introduction to using Model Context Protocol (MCP) to use with an AI Agent
• How to Build an Alternative to a RAG Agent That Provides Similar Functionality
• How To Connect This Agent With a Traditional RAG Agent (Using OpenSearch)

All of this using Small Language Models (SLM) which require less resources at inference time.

Software Prerequisites

• A Linux or Mac-based Developer’s Laptop with enough memory to run a database (ie OpenSearch) plus Intel's neural-chat SLM (below).
  • Windows Users should use a VM or Cloud Instance
• Python Installed: version 3.12 or higher
• (Recommended) Using a miniconda or venv virtual environment
• Docker (Linux or MacOS) Installed: for running a local OpenSearch instance
• Basic familiarity with shell operations

Docker images you should pre-pull in your environment:

• docker image pull opensearchproject/opensearch:3

LLM to pre-download:

This is the official one used in today's workshop:

• Intel's neural-chat-7B-v3-3-GGUF

OR

• Huggingface bartowski/Meta-Llama-3-8B-Instruct-GGUF

OR

• Alternatively, using ollama
  • Llama 3B: https://ollama.com/library/llama3:8b

Participation Options

There are 4 separate demos:

• demo/1_contradictory
• demo/2_mcp_whats_relevant
• demo/3_a2a_simple
• demo/4_a2a_financials

The instructions and purpose for each demo is contained within their respective folders.