AI agents are large language models (LLMs) enhanced with the ability to perform actions using external tools.
The Model Context Protocol (MCP) provides a standardized way for AI agents to connect with and use various tools, similar to REST APIs.
The Agent-to-Agent (A2A) protocol standardizes communication and collaboration between different AI agents.
The Strands SDK from AWS simplifies building multi-agent workflows by handling MCP and A2A protocols with just a few lines of code.
Learn how to instantiate a simple agent and configure it to use different LLM providers, including local models via Ollama.
Give agents new abilities by importing pre-built tools or creating your own custom functions using the `@tool` decorator.
Decouple tools from your agent by running them on a dedicated MCP server, making them accessible to any connected agent.
Create a collaborative system by running specialized agents on A2A servers and connecting them to a primary orchestrator agent.
A full demonstration shows a game master agent orchestrating a character agent, a rules agent, and a dice-rolling tool via a UI.
Find links to the Strands SDK documentation, code samples, and a discount code for AWS re:Invent.
1:32:50Chris Heilmann, Daniel Cranney & Tiffany Souterre
49:45Kris Howard
25:17Scott Hanselman
29:03Jörg Neumann
24:22Philipp Schmid
25:32Viktoria Semaan
26:35Ricardo
23:33Oliver Will
Jobs that call for the skills explored in this talk.
Deutsches Medizinrechenzentrum GmbH
Gemma Analytics
4Com GmbH & Co. KG