Streamlit + LangChain RAG Chatbot 2026

Build a private, offline RAG (Retrieval-Augmented Generation) chatbot that answers questions based on your documents — using Streamlit + LangChain + Ollama in 2026.

Why Build This in 2026?

• Full privacy – documents never leave your machine
• No API costs after setup (local Llama 3 / Mistral)
• Streamlit = fastest way to build interactive AI apps
• LangChain + Ollama = powerful local RAG stack
• Very popular starter project – high search volume

Prerequisites

• Python 3.11+ (3.14 recommended)
• Ollama installed – https://ollama.com/download
• Model: ollama pull llama3.1:8b (or mistral, phi3, gemma2)
• pip install streamlit langchain langchain-community langchain-ollama chromadb pypdf

1. Basic RAG Chatbot Setup

# app.py
import streamlit as st
from langchain_ollama import OllamaEmbeddings, ChatOllama
from langchain_community.vectorstores import Chroma
from langchain_community.document_loaders import PyPDFLoader
from langchain_text_splitters import RecursiveCharacterTextSplitter
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.runnables import RunnablePassthrough
from langchain_core.output_parsers import StrOutputParser

# LLM & Embeddings
llm = ChatOllama(model="llama3.1:8b")
embeddings = OllamaEmbeddings(model="llama3.1:8b")

# Prompt template
prompt = ChatPromptTemplate.from_template(
    """Answer the question based only on the following context:
{context}

Question: {question}
"""
)

# RAG chain
rag_chain = (
    {"context": retriever | format_docs, "question": RunnablePassthrough()}
    | prompt
    | llm
    | StrOutputParser()
)

# Streamlit UI
st.title("Private RAG Chatbot 2026")
uploaded_file = st.file_uploader("Upload PDF", type="pdf")

if uploaded_file:
    # Save & load PDF
    with open("temp.pdf", "wb") as f:
        f.write(uploaded_file.getbuffer())
    
    loader = PyPDFLoader("temp.pdf")
    docs = loader.load()
    
    # Split & embed
    text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
    splits = text_splitter.split_documents(docs)
    vectorstore = Chroma.from_documents(documents=splits, embedding=embeddings)
    retriever = vectorstore.as_retriever(search_kwargs={"k": 4})

st.header("Ask questions about the document")
question = st.text_input("Your question:")

if question:
    response = rag_chain.invoke(question)
    st.write("**Answer:**", response)
        

2. Add Memory & Tools (Advanced)

from langchain.memory import ConversationBufferMemory
from langchain.chains import ConversationalRetrievalChain

memory = ConversationBufferMemory(memory_key="chat_history", return_messages=True)

qa_chain = ConversationalRetrievalChain.from_llm(
    llm=llm,
    retriever=retriever,
    memory=memory
)

# In Streamlit
if "messages" not in st.session_state:
    st.session_state.messages = []

for message in st.session_state.messages:
    with st.chat_message(message["role"]):
        st.markdown(message["content"])

if prompt := st.chat_input("Ask anything about the document"):
    st.session_state.messages.append({"role": "user", "content": prompt})
    with st.chat_message("user"):
        st.markdown(prompt)

    with st.chat_message("assistant"):
        response = qa_chain({"question": prompt})
        st.markdown(response["answer"])
    st.session_state.messages.append({"role": "assistant", "content": response["answer"]})
        

3. Deployment & Scaling 2026

• Local: streamlit run app.py
• Docker: Add Dockerfile + EXPOSE 8501
• Cloud: Railway / Fly.io / Render (free tier)
• Secure: Add login (streamlit-authenticator)
• Scale: Use multiple Ollama instances + nginx load balancer