Agentic RAG with LangChain, FAISS, and Ollama

pip install -U langchain langchain-ollama langchain-community langchain-core faiss-cpu
ollama pull nomic-embed-text
ollama pull qwen3

llm = ChatOllama(
    model="qwen3",
    base_url="http://localhost:11434"
)

embeddings = OllamaEmbeddings(
    model="nomic-embed-text",
    base_url="http://localhost:11434"
)

db_name = "./../09. Vector Stores and Retrievals/health_supplements"
vector_store = FAISS.load_local(
    db_name,
    embeddings,
    allow_dangerous_deserialization=True
)

llm.invoke("Hello")

AIMessage(content='Hello! How can I assist you today? 😊', ...)

@tool()
def retrieve_context(query: str):
    """Retrieve relevant information for health related queries from the document to answer the query.

    """
    print(f"🔍 Searching: '{query}'")

    docs = vector_store.similarity_search(query, k=4)

    content = "\n\n".join(
        f"Source: {doc.metadata.get('source', '?')} (Page {doc.metadata.get('page', '?')}): {doc.page_content}"
        for doc in docs
    )

    print(f"✔ Found {len(docs)} relevant chunks")
    return content

tools = [retrieve_context]

system_prompt = """You are a research assistant with a document retrieval tool.

                    Tool:
                    - retrieve_context: Search the document for the health related question

                    Cite page numbers and reference document while writing the answer and be thorough."""

rag_agent = create_agent(llm, tools, system_prompt=system_prompt)

result

{'messages': [
  HumanMessage(content='What is the use of BCAA?', ...),
  AIMessage(content='', ..., tool_calls=[{'name': 'retrieve_context', 'args': {'query': 'use of BCAA'}, ...}], ...),
  ToolMessage(content='Source: rag-dataset\\gym supplements\\1. Analysis of Actual Fitness Supplement.pdf (Page 1):
    Foods 2024, 13, 1424\n2 of 21\nand sports industry, evidence suggests that creatine can benefit not only athletes
    but also the elderly and the general population [6]. Branched-chain amino acids (BCAA) also offer a plethora
    of benefits for consumers. As explained by Sanz et al. [7], BCAAs are stored directly in muscles and serve
    as the raw materials needed to build new muscle...
    \n\nSource: rag-dataset\\health supplements\\3.health_supplements_side_effects.pdf (Page 7):
    ...DMAA-containing supplements...', name='retrieve_context', ...),
  AIMessage(content='Branched-Chain Amino Acids (BCAA) are primarily used for their role in muscle metabolism and recovery...', ...)
]}

def ask(question: str):
    """Ask the agentic RAG a question."""
    print(f"\n{'='*60}")
    print(f"Question: {question}")
    print('='*60)

    for event in rag_agent.stream(
        {"messages": [{"role": "user", "content": question}]},
        stream_mode="values"
    ):
        msg = event["messages"][-1]

        # Show tool usage
        if hasattr(msg, 'tool_calls') and msg.tool_calls:
            for tc in msg.tool_calls:
                print("Tool Call: ")
                print(f"\n🔺 Using: {tc['name']} with {tc['args']}")

        # Show final answer
        elif hasattr(msg, 'content') and msg.content:
            print(f"\n💼 Answer:\n{msg.content}")

def chat():
    """Start interactive chat with the agentic RAG."""
    print("\n🤖 Agentic RAG Chat - Type 'quit', 'q', or 'exit' to exit")

    while True:
        question = input("\nYour question: ").strip()
        if question.lower() in ['quit', 'exit', 'q']:
            break
        if question:
            ask(question)

chat()

🤖 Agentic RAG Chat - Type 'quit, q or exit' to exit

============================================================
Question: tell me about sun?
============================================================

💼 Answer:
The sun is a star at the center of our solar system, composed primarily of hydrogen (about 75%)
and helium (about 25%)... If you were asking about health-related aspects of the sun
(e.g., sunlight's role in vitamin D synthesis or skin cancer risks), I could retrieve specific
health information. Let me know if you'd like to focus on a specific aspect!

============================================================
Question: tell me about the protein?
============================================================

💼 Answer:
tell me about the protein?
Tool Call:

🔺 Using: retrieve_context with {'query': 'protein'}
🔍 Searching: 'protein'
✔ Found 4 relevant chunks

💼 Answer:
Source: rag-dataset\health supplements\3.health_supplements_side_effects.pdf (Page 3):
PROTEIN POWDERS AND INFANT FORMULA
Protein powders consisting of the dairy proteins casein, whey and of vegetable proteins in soy
protein isolate (SPI) are popular supplements among athletes and body builders...