Create app.py

app.py

@@ -0,0 +1,80 @@
+import os
+import gradio as gr
+from autogen import AssistantAgent
+from autogen.agentchat.contrib.retrieve_user_proxy_agent import RetrieveUserProxyAgent
+from dotenv import load_dotenv
+
+load_dotenv()  # Load environment variables
+
+gpt4 = {
+    "config_list": [{"model": "gpt-4", "api_key": os.environ["OPENAI_API_KEY"]}],
+}
+
+# Initialize agents
+assistant = AssistantAgent(
+    name="assistant",
+    system_message="You are a helpful assistant. Give the user information with a set of logical steps to follow. In addition provide helpful url resources from ragproxyagent.",
+    llm_config=gpt4,
+)
+
+ragproxyagent = RetrieveUserProxyAgent(
+    name="ragproxyagent",
+    human_input_mode="ALWAYS",
+    retrieve_config={
+        "task": "qa",
+        "docs_path": "./scraped_results/",
+        "context_max_tokens": 5000,
+        "overwrite": False,
+        "get_or_create": True,
+        "return_source": True,
+    },
+    code_execution_config=False,
+)
+
+# Function to reset agents
+def _reset_agents():
+    ragproxyagent.reset()
+    assistant.reset()
+
+import time
+
+
+def chat_with_agents(user_input, history):
+    _reset_agents()  # Reset agents before each new conversation
+
+    # Step 1: RAG Agent retrieves context (doc-based information)
+    chat_result = ragproxyagent.initiate_chat(
+        assistant,
+        message=user_input,
+        problem=user_input,
+        n_results=3,
+        max_turns=1,
+    )
+
+    print("Raw RAG Response:", chat_result)  # Debugging
+
+    # Step 2: Extract assistant's final summarized response
+    response = "Sorry, I couldn't generate a response."
+
+    # Check if the response contains the summary
+    if hasattr(chat_result, 'summary') and chat_result.summary:
+        response = chat_result.summary
+
+   # print("Assistant's Final Response:", response)  # Debugging
+
+    # Step 3: Append user and assistant messages
+   # history.append({"role": "user", "content": user_input})
+   # history.append({"role": "assistant", "content": response})
+
+    # Return only the last 2 messages (user + assistant) for Gradio
+    return response
+
+# Use "messages" type for proper chat UI
+chat_interface = gr.ChatInterface(
+    fn=chat_with_agents,
+    type="messages"
+)
+
+if __name__ == "__main__":
+    chat_interface.launch()
+