diff --git a/core/cat/factory/custom_llm.py b/core/cat/factory/custom_llm.py
index b909968d..d1104c32 100644
--- a/core/cat/factory/custom_llm.py
+++ b/core/cat/factory/custom_llm.py
@@ -14,6 +14,10 @@ def _llm_type(self):
     def _call(self, prompt, stop=None):
         return "AI: You did not configure a Language Model. " \
                "Do it in the settings!"
+    
+    async def _acall(self, prompt, stop=None):
+        return "AI: You did not configure a Language Model. " \
+               "Do it in the settings!"
 
 
 # elaborated from
diff --git a/core/cat/looking_glass/agent_manager.py b/core/cat/looking_glass/agent_manager.py
index b6810de7..4e5dcc08 100644
--- a/core/cat/looking_glass/agent_manager.py
+++ b/core/cat/looking_glass/agent_manager.py
@@ -32,7 +32,7 @@ def __init__(self, cat):
         self.cat = cat
 
 
-    def execute_tool_agent(self, agent_input, allowed_tools):
+    async def execute_tool_agent(self, agent_input, allowed_tools):
 
         allowed_tools_names = [t.name for t in allowed_tools]
         # TODO: dynamic input_variables as in the main prompt 
@@ -69,11 +69,11 @@ def execute_tool_agent(self, agent_input, allowed_tools):
             verbose=True
         )
 
-        out = agent_executor(agent_input)
+        out = await agent_executor.acall(agent_input)
         return out
     
 
-    def execute_memory_chain(self, agent_input, prompt_prefix, prompt_suffix, working_memory: WorkingMemory):
+    async def execute_memory_chain(self, agent_input, prompt_prefix, prompt_suffix, working_memory: WorkingMemory):
 
         input_variables = [i for i in agent_input.keys() if i in prompt_prefix + prompt_suffix]
         # memory chain (second step)
@@ -88,13 +88,13 @@ def execute_memory_chain(self, agent_input, prompt_prefix, prompt_suffix, workin
             verbose=True
         )
 
-        out = memory_chain(agent_input, callbacks=[NewTokenHandler(self.cat, working_memory)])
+        out = await memory_chain.acall(agent_input, callbacks=[NewTokenHandler(self.cat, working_memory)])
         out["output"] = out["text"]
         del out["text"]
         return out
 
 
-    def execute_agent(self, working_memory):
+    async def execute_agent(self, working_memory):
         """Instantiate the Agent with tools.
 
         The method formats the main prompt and gather the allowed tools. It also instantiates a conversational Agent
@@ -134,7 +134,7 @@ def execute_agent(self, working_memory):
             log.debug(f"{len(allowed_tools)} allowed tools retrived.")
 
             try:
-                tools_result = self.execute_tool_agent(agent_input, allowed_tools)
+                tools_result = await self.execute_tool_agent(agent_input, allowed_tools)
 
                 # If tools_result["output"] is None the LLM has used the fake tool none_of_the_others  
                 # so no relevant information has been obtained from the tools.
@@ -177,7 +177,7 @@ def execute_agent(self, working_memory):
         #Adding the tools_output key in agent input, needed by the memory chain
         agent_input["tools_output"] = ""
         # Execute the memory chain
-        out = self.execute_memory_chain(agent_input, prompt_prefix, prompt_suffix, working_memory)
+        out = await self.execute_memory_chain(agent_input, prompt_prefix, prompt_suffix, working_memory)
 
         return out
     
diff --git a/core/cat/looking_glass/cheshire_cat.py b/core/cat/looking_glass/cheshire_cat.py
index 7a633676..53bc6573 100644
--- a/core/cat/looking_glass/cheshire_cat.py
+++ b/core/cat/looking_glass/cheshire_cat.py
@@ -398,7 +398,7 @@ def send_ws_message(self, content: str, msg_type: MSG_TYPES = "notification", wo
                 )
             )
 
-    def __call__(self, user_message_json):
+    async def __call__(self, user_message_json):
         """Call the Cat instance.
 
         This method is called on the user's message received from the client.
@@ -456,7 +456,7 @@ def __call__(self, user_message_json):
         
         # reply with agent
         try:
-            cat_message = self.agent_manager.execute_agent(user_working_memory)
+            cat_message = await self.agent_manager.execute_agent(user_working_memory)
         except Exception as e:
             # This error happens when the LLM
             #   does not respect prompt instructions.
diff --git a/core/cat/routes/websocket.py b/core/cat/routes/websocket.py
index 5e2a828b..1cb0998f 100644
--- a/core/cat/routes/websocket.py
+++ b/core/cat/routes/websocket.py
@@ -62,7 +62,7 @@ async def receive_message(ccat: CheshireCat, user_id: str = "user"):
         user_message["user_id"] = user_id
 
         # Run the `ccat` object's method in a threadpool since it might be a CPU-bound operation.
-        cat_message = await run_in_threadpool(ccat, user_message)
+        cat_message = await ccat(user_message)
 
         # Send the response message back to the user.
         await manager.send_personal_message(cat_message, user_id)