# Initialize LLM
    self.llm = ChatOpenAI(model_name="gpt-3.5-turbo", temperature=0)
    
    # Initialize vector store
    self.vector_store = None
    
    # Text splitter for long documents
    self.text_splitter = RecursiveCharacterTextSplitter(
        chunk_size=500,
        chunk_overlap=50
    )
    
    # Search prompt
    self.search_prompt = ChatPromptTemplate.from_template("""
    You are a helpful course recommendation assistant. Based on the following course information 
    and the user's query, provide a concise response recommending the most relevant courses.
    
    Course Information:
    {context}
    
    User Query: {query}
    
    Please explain why these courses match their needs and include any relevant details about difficulty 
    level and duration. Keep your response clear and helpful.
    """)
    
    # Create the chain
    self.chain = (
        self.search_prompt 
        | self.llm 
        | StrOutputParser()
    )

def load_courses(self, courses: List[Course]):
    """Load courses into the vector store"""
    texts = []
    metadatas = []
    
    for course in courses:
        # Combine course info into a single string
        course_text = f"""
        Title: {course.title}
        Description: {course.description}
        Curriculum: {course.curriculum}
        Difficulty: {course.difficulty_level}
        Duration: {course.duration}
        """
        
        # Split long texts
        splits = self.text_splitter.split_text(course_text)
        texts.extend(splits)
        
        # Add metadata for each split
        for _ in splits:
            metadatas.append({
                "title": course.title,
                "difficulty": course.difficulty_level,
                "duration": course.duration
            })
    
    # Create vector store
    self.vector_store = FAISS.from_texts(
        texts,
        self.embed_model,
        metadatas=metadatas
    )

def search(self, query: str, k: int = 3) -> str:
    """Search for courses based on the query"""
    if not self.vector_store:
        return "No courses loaded in the system."
    
    # Retrieve relevant documents
    docs = self.vector_store.similarity_search(query, k=k)
    
    # Combine context from retrieved documents
    context = "\n\n".join(doc.page_content for doc in docs)
    
    # Generate response using the chain
    response = self.chain.invoke({
        "context": context,
        "query": query
    })
    
    return response

interface = gr.Interface(
    fn=search_courses,
    inputs=gr.Textbox(
        lines=2,
        placeholder="Enter your search query (e.g., 'beginner python course' or 'advanced machine learning')"
    ),
    outputs=gr.Textbox(lines=10),
    title="Smart Course Search",
    description="Search for relevant courses based on your interests and requirements.",
    examples=[
        ["I want to learn Python as a beginner"],
        ["Show me advanced machine learning courses"],
        ["What courses cover data visualization?"]
    ]
)
return interface