<ul><li>LLMs, while powerful, have limitations in answering questions based on their training knowledge.</li><li>Agentic frameworks have expanded LLM functionalities, including reasoning, internet search, and deep research.</li><li>Agents are workflows around LLMs that make choices around tools needed for specific tasks.</li><li>Langgraph is a popular agentic framework that enables creating workflows around LLMs.</li><li>MCP protocol by Anthropic connects AI models to different data sources and tools across the Internet.</li><li>MCP allows clients to access agent components together in a standardized and secure manner.</li><li>A2A protocol focuses on agent collaboration and complements MCP for connecting sub-agents.</li><li>Both MCP and A2A frameworks use Starlette for web services but differ in protocol methods and formats.</li><li>MCP and A2A enable easy setup, communication, and resource access for agents, aiding in integration of LLMs.</li><li>Open-source MCP and A2A facilitate organizations to build and share agents, enabling seamless communication across systems.</li></ul>

Agentic Frameworks and Protocols

Discover more