diff --git a/backend/modules/brain/integrations/Big/Brain.py b/backend/modules/brain/integrations/Big/Brain.py
index ab7cd9858..bf210a45c 100644
--- a/backend/modules/brain/integrations/Big/Brain.py
+++ b/backend/modules/brain/integrations/Big/Brain.py
@@ -36,7 +36,8 @@ class BigBrain(KnowledgeBrainQA):
 
     def get_chain(self):
         system_template = """Combine these summaries in a way that makes sense and answer the user's question.
-        Use markdown or any other techniques to display the content in a nice and aerated way. 
+        Use markdown or any other techniques to display the content in a nice and aerated way.
+        Here are user instructions on how to respond: {custom_personality}
         ______________________
         {summaries}"""
         messages = [
@@ -88,6 +89,9 @@ class BigBrain(KnowledgeBrainQA):
             {
                 "question": question.question,
                 "chat_history": transformed_history,
+                "custom_personality": (
+                    self.prompt_to_use.content if self.prompt_to_use else None
+                ),
             }
         ):
             if "answer" in chunk:
diff --git a/backend/modules/brain/integrations/GPT4/Brain.py b/backend/modules/brain/integrations/GPT4/Brain.py
index 03db2cd1d..505b4d4f9 100644
--- a/backend/modules/brain/integrations/GPT4/Brain.py
+++ b/backend/modules/brain/integrations/GPT4/Brain.py
@@ -28,7 +28,10 @@ class GPT4Brain(KnowledgeBrainQA):
 
         prompt = ChatPromptTemplate.from_messages(
             [
-                ("system", "You are GPT-4 powered by Quivr. You are an assistant."),
+                (
+                    "system",
+                    "You are GPT-4 powered by Quivr. You are an assistant. {custom_personality}",
+                ),
                 MessagesPlaceholder(variable_name="chat_history"),
                 ("human", "{question}"),
             ]
@@ -53,6 +56,9 @@ class GPT4Brain(KnowledgeBrainQA):
             {
                 "question": question.question,
                 "chat_history": transformed_history,
+                "custom_personality": (
+                    self.prompt_to_use.content if self.prompt_to_use else None
+                ),
             }
         ):
             response_tokens.append(chunk.content)