Update app.py

app.py

@@ -15,6 +15,8 @@ from langchain_groq import ChatGroq
 from langchain_huggingface import HuggingFaceEndpoint, ChatHuggingFace
 from langchain_huggingface.llms import HuggingFacePipeline
 from langchain_ollama import ChatOllama
+from langchain_google_genai import ChatGoogleGenerativeAI
+from langchain_core.rate_limiters import InMemoryRateLimiter
 from smolagents import (
     InferenceClientModel, LiteLLMModel, OpenAIServerModel, TransformersModel,
     CodeAgent,
@@ -64,28 +66,34 @@ HF_TOKEN = os.getenv("HF_TOKEN")
 #chat = ChatHuggingFace(llm=llm, verbose=True)
 #chat = ChatOllama(llm=hf_pipe).bind(skip_prompt=True)
 
-#tools = []
-#chat_with_tools = chat.bind_tools(tools)
+#openai_api_key = os.getenv("OPENAI_API_KEY")
 
-openai_api_key = os.getenv("OPENAI_API_KEY")
-
-model = OpenAIServerModel(
-    api_key=openai_api_key,
-    model_id="gpt-4.1"
+#model = OpenAIServerModel(
+#    api_key=openai_api_key,
+#    model_id="gpt-4.1"
+#)
+#tools = [
+#    DuckDuckGoSearchTool(),
+#    PythonInterpreterTool(),
+#]
+
+rate_limiter = InMemoryRateLimiter(
+    # <-- Super slow! We can only make a request once every 4 seconds!!
+    requests_per_second=15/60,
+    # Wake up every 100 ms to check whether allowed to make a request,
+    check_every_n_seconds=0.1,
+    max_bucket_size=10,  # Controls the maximum burst size.
 )
-tools = [
-    DuckDuckGoSearchTool(),
-    PythonInterpreterTool(),
-]
+
+chat = ChatGoogleGenerativeAI(model="gemini-2.0-flash", temperature=0, rate_limiter=rate_limiter)
+tools = []
+chat_with_tools = chat.bind_tools(tools)
 
 # load the system prompt from the file
 with open("system_prompt.txt", "r", encoding="utf-8") as f:
     system_prompt = f.read()
 
 print(system_prompt)
-
-# System message
-sys_msg = SystemMessage(content=system_prompt)
     
 def assistant(state: MessagesState):
     return {
@@ -106,7 +114,9 @@ class BasicAgent:
         
     def __call__(self, question: str) -> str:
         print(f"Agent received question (first 50 chars): {question[:50]}...")
-        messages = [HumanMessage(content=question)]
+        messages = [
+            SystemMessage(content=system_prompt),
+            HumanMessage(content=question)]
         response = self.graph.invoke({"messages": messages})
         print(f"RESPONSE {response}")
         response = response['messages'][-1].content