Agentic AI

Provides components for building agents:

1. Model wrappers

   Wrapper for providers like OpenAI, Anthropic, Local, etc.

2. Message types
3. Prompt Templates

4. Tools

   You can convert any python function to a tool just by decorating it with @tools

5. Structured Output

6. Document Loaders

   For ingesting data from sources like PDFs, websites, Markdown files, Google Drive, Notion, S3, etc. PyPDFLoader

7. Text Splitters: RecursiveCharacterTextSplitter
8. Embedding model wrappers
9. Vector store integrations
10. Retrievers: It can use the vector store or can use external APIs

11. Agents

    from langchain.agents import create_agent
    
    agent = create_agent(
        model=llm,
        tools=[get_weather, search_docs],
        system_prompt="You are a helpful assistant."
    )
    

12. Chains

LangGraph provides a graph based state transition abstraction, persistence for those states, and other useful tooling.