diff --git a/llm_demo/orchestrator/langchain_tools/langchain_tools_orchestrator.py b/llm_demo/orchestrator/langchain_tools/langchain_tools_orchestrator.py
index 1595b052..f1ee7679 100644
--- a/llm_demo/orchestrator/langchain_tools/langchain_tools_orchestrator.py
+++ b/llm_demo/orchestrator/langchain_tools/langchain_tools_orchestrator.py
@@ -69,8 +69,8 @@ def initialize_agent(
         prompt: ChatPromptTemplate,
         model: str,
     ) -> "UserAgent":
+        # TODO: Use .bind_tools(tools) to bind the tools with the LLM.
         llm = ChatVertexAI(max_output_tokens=512, model_name=model, temperature=0.0)
-        llm.bind_tools(tools)
         memory = ConversationBufferMemory(
             chat_memory=ChatMessageHistory(messages=history),
             memory_key="chat_history",
diff --git a/llm_demo/orchestrator/langgraph/react_graph.py b/llm_demo/orchestrator/langgraph/react_graph.py
index 3472ee5a..b05134f1 100644
--- a/llm_demo/orchestrator/langgraph/react_graph.py
+++ b/llm_demo/orchestrator/langgraph/react_graph.py
@@ -85,6 +85,7 @@ async def create_graph(
     tool_node = ToolNode(tools)
 
     # model node
+    # TODO: Use .bind_tools(tools) to bind the tools with the LLM.
     model = ChatVertexAI(max_output_tokens=512, model_name=model_name, temperature=0.0)
 
     # Add the prompt to the model to create a model runnable
@@ -98,6 +99,10 @@ async def acall_model(state: UserState, config: RunnableConfig):
         messages = state["messages"]
         res = await model_runnable.ainvoke({"messages": messages}, config)
 
+        # TODO: Remove the temporary fix of parsing LLM response and invoking
+        # tools until we use bind_tools API and have automatic response parsing
+        # and tool calling. (see
+        # https://langchain-ai.github.io/langgraph/#example)
         if "```json" in res.content:
             try:
                 response = str(res.content).replace("```json", "").replace("```", "")