Building a Fuji X-S20 Q&A App with Gemini, LangChain and Gradio

Navigation

• Intro
• What You’ll Build
• Understanding RAG
• LangChain Overview
• Step-by-Step Build
• Run the App
• Conclusion

---

Intro

In this tutorial, we build a Fuji X-S20 Question Answering App using:

• Gemini API
• LangChain
• Gradio

The goal is simple:

👉 Ask questions about your camera without reading a 400-page manual

---

What You’ll Build

A RAG-powered app that:

• reads a PDF manual
• retrieves relevant context
• answers user questions

---

Understanding RAG

RAG (Retrieval Augmented Generation) works in 4 steps:

1. Load data
2. Split into chunks
3. Create embeddings
4. Store in vector database

---

📸 RAG Flow

---

LangChain Overview

LangChain helps:

• manage embeddings
• connect to vector databases
• build retrieval pipelines

---

📸 LangChain Flow

---

Step-by-Step Build

Step 1: Environment Setup

import os
import gradio as gr
from dotenv import load_dotenv
from langchain_community.document_loaders import PyPDFLoader
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain_chroma import Chroma
from langchain_google_genai import GoogleGenerativeAIEmbeddings
from langchain_google_genai import ChatGoogleGenerativeAI
from langchain.chains import create_retrieval_chain
from langchain.chains.combine_documents import create_stuff_documents_chain
from langchain_core.prompts import ChatPromptTemplate

---

Step 2: Load Environment Variables

load_dotenv()
api_key = os.getenv("API_KEY")

---

Step 3: Load PDF Manual

loader = PyPDFLoader("Fuji_xs20_manual.pdf")
data = loader.load()

---

Step 4: Split Text

text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000)
docs = text_splitter.split_documents(data)

---

Step 5: Create Embeddings

embeddings = GoogleGenerativeAIEmbeddings(api_key=api_key, model="models/embedding-001")
vectorstore = Chroma.from_documents(documents=docs, embedding=embeddings)
retriever = vectorstore.as_retriever(search_type="similarity", search_kwargs={"k": 10})

---

Step 6: Configure LLM

llm = ChatGoogleGenerativeAI(model="gemini-1.5-pro", temperature=0.3, max_tokens=500)

---

Step 7: Prompt Template

system_prompt = (
    "You are an assistant for question-answering tasks. "
    "Use retrieved context to answer. If unknown, say you don't know. "
    "Keep answers concise."
)

prompt = ChatPromptTemplate.from_messages([
    ("system", system_prompt),
    ("human", "{input}")
])

---

Step 8: Create RAG Chain

qa_chain = create_stuff_documents_chain(llm, prompt)
rag_chain = create_retrieval_chain(retriever, qa_chain)

---

Step 9: Query Function

def answer_query(query):
    if query:
        response = rag_chain.invoke({"input": query})
        return response["answer"]

---

Step 10: Gradio Interface

iface = gr.Interface(
    fn=answer_query,
    inputs="text",
    outputs="text",
    title="Fuji X-S20 Q&A App",
    description="Ask any question about your Fuji camera"
)

if __name__ == "__main__":
    iface.launch()

---

Run the App

python app.py

---

📸 App Interface

---

Conclusion

You now have a working RAG application that:

• understands documents
• retrieves context
• answers questions

---

Key Insight

RAG allows AI systems to answer questions using your own data

---

📩 Subscribe

If you’re building AI systems:

I share practical workflows, tools, and experiments every time I publish.