Update app.py

app.py

@@ -7,7 +7,7 @@ faq_df = pd.read_csv("lic_faq.csv")
 model = SentenceTransformer('all-MiniLM-L6-v2')
 faq_embeddings = model.encode(faq_df['question'].tolist(), convert_to_tensor=True)
 
-# Sample policies to suggest (add more as needed)
+# Policy suggestion mapping
 policy_suggestions = {
     "term": "💡 You might consider LIC Tech Term Plan for pure protection at low cost.",
     "money back": "💡 LIC Money Back Policy is great for periodic returns along with insurance.",
@@ -16,28 +16,35 @@ policy_suggestions = {
     "pension": "💡 LIC Jeevan Akshay and PM Vaya Vandana Yojana are best for pension seekers."
 }
 
+# Handle chat
 def chatbot(history, query):
-    query_embedding = model.encode(query, convert_to_tensor=True)
-    scores = util.pytorch_cos_sim(query_embedding, faq_embeddings)[0]
-    best_score = float(scores.max())
-    best_idx = int(scores.argmax())
+    query_lower = query.lower().strip()
 
-    if best_score < 0.6:
-        response = "🤖 I'm not confident I have the right answer for that. Please ask about LIC policies, claims, commissions, onboarding, or KYC."
+    # Handle greetings
+    if query_lower in ["hi", "hello", "hey", "good morning", "good evening"]:
+        response = "👋 Hello! I’m your LIC Assistant. Ask me anything about LIC policies, claims, onboarding, or commission."
     else:
-        response = faq_df.iloc[best_idx]['answer']
+        # Embedding + similarity
+        query_embedding = model.encode(query, convert_to_tensor=True)
+        scores = util.pytorch_cos_sim(query_embedding, faq_embeddings)[0]
+        best_score = float(scores.max())
+        best_idx = int(scores.argmax())
 
-    # Policy recommendation based on keywords
-    query_lower = query.lower()
-    for keyword, suggestion in policy_suggestions.items():
-        if keyword in query_lower:
-            response += f"\n\n{suggestion}"
-            break
+        if best_score < 0.6:
+            response = "🤖 I'm not confident I have the right answer for that. Please ask about LIC policies, claims, commissions, onboarding, or KYC."
+        else:
+            response = faq_df.iloc[best_idx]['answer']
+
+        # Policy suggestion
+        for keyword, suggestion in policy_suggestions.items():
+            if keyword in query_lower:
+                response += f"\n\n{suggestion}"
+                break
 
     history.append((query, response))
     return history, history
 
-# UI using Gradio Blocks
+# Gradio UI
 with gr.Blocks(title="LIC Agent Chatbot") as demo:
     gr.Markdown(
         "<h1 style='text-align:center;color:#0D47A1;'>🧑‍💼 LIC Agent Assistant</h1>"
@@ -45,12 +52,14 @@ with gr.Blocks(title="LIC Agent Chatbot") as demo:
     )
 
     chatbot_ui = gr.Chatbot(label="LIC Assistant", height=450)
-    msg = gr.Textbox(label="Your Question", placeholder="E.g., What is the commission for ULIP?")
+    with gr.Row():
+        msg = gr.Textbox(placeholder="E.g., What is the commission for ULIP?", label="Your Question", scale=8)
+        send = gr.Button("Send", variant="primary", scale=2)
     clear = gr.Button("Clear Chat")
 
     state = gr.State([])
 
-    msg.submit(chatbot, [state, msg], [chatbot_ui, state])
+    send.click(fn=chatbot, inputs=[state, msg], outputs=[chatbot_ui, state])
     clear.click(lambda: ([], []), None, [chatbot_ui, state], queue=False)
 
 demo.launch()