diff --git a/app.py b/app.py
index 0740b42..3dbebf0 100644
--- a/app.py
+++ b/app.py
@@ -27,6 +27,7 @@ cred = credentials.Certificate(os.getenv("GOOGLE_APPLICATION_CREDENTIALS"))
 firebase_admin.initialize_app(cred)
 
 GRADING_TEMPERATURE = 0.1
+TIPS_TEMPERATURE = 0.2
 GEN_QUESTION_TEMPERATURE = 0.9
 GPT_3_5_TURBO = "gpt-3.5-turbo"
 GPT_3_5_TURBO_16K = "gpt-3.5-turbo-16k"
@@ -38,6 +39,7 @@ FIREBASE_BUCKET = 'mti-ielts.appspot.com'
 AUDIO_FILES_PATH = 'download-audio/'
 FIREBASE_LISTENING_AUDIO_FILES_PATH = 'listening_recordings/'
 
+
 @app.route('/listening_section_1', methods=['GET'])
 @jwt_required()
 def get_listening_section_1_question():
@@ -45,11 +47,13 @@ def get_listening_section_1_question():
         messages = get_question_gen_messages(QuestionType.LISTENING_SECTION_1)
         token_count = reduce(lambda count, item: count + count_tokens(item)['n_tokens'],
                              map(lambda x: x["content"], filter(lambda x: "content" in x, messages)), 0)
-        response = make_openai_call(GPT_3_5_TURBO_16K, messages, token_count, LISTENING_GEN_FIELDS, GEN_QUESTION_TEMPERATURE)
+        response = make_openai_call(GPT_3_5_TURBO_16K, messages, token_count, LISTENING_GEN_FIELDS,
+                                    GEN_QUESTION_TEMPERATURE)
         return response
     except Exception as e:
         return str(e)
 
+
 @app.route('/listening_section_2', methods=['GET'])
 @jwt_required()
 def get_listening_section_2_question():
@@ -58,7 +62,8 @@ def get_listening_section_2_question():
         messages = get_question_gen_messages(QuestionType.LISTENING_SECTION_2)
         token_count = reduce(lambda count, item: count + count_tokens(item)['n_tokens'],
                              map(lambda x: x["content"], filter(lambda x: "content" in x, messages)), 0)
-        response = make_openai_call(GPT_3_5_TURBO_16K, messages, token_count, LISTENING_GEN_FIELDS, GEN_QUESTION_TEMPERATURE)
+        response = make_openai_call(GPT_3_5_TURBO_16K, messages, token_count, LISTENING_GEN_FIELDS,
+                                    GEN_QUESTION_TEMPERATURE)
 
         # file_name = str(uuid.uuid4()) + ".mp3"
         # sound_file_path = AUDIO_FILES_PATH + file_name
@@ -70,6 +75,7 @@ def get_listening_section_2_question():
     except Exception as e:
         return str(e)
 
+
 @app.route('/listening_section_3', methods=['GET'])
 @jwt_required()
 def get_listening_section_3_question():
@@ -78,7 +84,8 @@ def get_listening_section_3_question():
         messages = get_question_gen_messages(QuestionType.LISTENING_SECTION_3)
         token_count = reduce(lambda count, item: count + count_tokens(item)['n_tokens'],
                              map(lambda x: x["content"], filter(lambda x: "content" in x, messages)), 0)
-        response = make_openai_call(GPT_3_5_TURBO_16K, messages, token_count, LISTENING_GEN_FIELDS, GEN_QUESTION_TEMPERATURE)
+        response = make_openai_call(GPT_3_5_TURBO_16K, messages, token_count, LISTENING_GEN_FIELDS,
+                                    GEN_QUESTION_TEMPERATURE)
 
         # file_name = str(uuid.uuid4()) + ".mp3"
         # sound_file_path = AUDIO_FILES_PATH + file_name
@@ -90,6 +97,7 @@ def get_listening_section_3_question():
     except Exception as e:
         return str(e)
 
+
 @app.route('/listening_section_4', methods=['GET'])
 @jwt_required()
 def get_listening_section_4_question():
@@ -98,7 +106,8 @@ def get_listening_section_4_question():
         messages = get_question_gen_messages(QuestionType.LISTENING_SECTION_4)
         token_count = reduce(lambda count, item: count + count_tokens(item)['n_tokens'],
                              map(lambda x: x["content"], filter(lambda x: "content" in x, messages)), 0)
-        response = make_openai_call(GPT_3_5_TURBO_16K, messages, token_count, LISTENING_GEN_FIELDS, GEN_QUESTION_TEMPERATURE)
+        response = make_openai_call(GPT_3_5_TURBO_16K, messages, token_count, LISTENING_GEN_FIELDS,
+                                    GEN_QUESTION_TEMPERATURE)
 
         # file_name = str(uuid.uuid4()) + ".mp3"
         # sound_file_path = AUDIO_FILES_PATH + file_name
@@ -110,6 +119,7 @@ def get_listening_section_4_question():
     except Exception as e:
         return str(e)
 
+
 @app.route('/writing_task1', methods=['POST'])
 @jwt_required()
 def grade_writing_task_1():
@@ -126,6 +136,7 @@ def grade_writing_task_1():
     except Exception as e:
         return str(e)
 
+
 @app.route('/writing_task2', methods=['POST'])
 @jwt_required()
 def grade_writing_task_2():
@@ -141,6 +152,7 @@ def grade_writing_task_2():
     except Exception as e:
         return str(e)
 
+
 @app.route('/fetch_tips', methods=['POST'])
 @jwt_required()
 def fetch_answer_tips():
@@ -153,7 +165,7 @@ def fetch_answer_tips():
         messages = get_question_tips(question, answer, correct_answer, context)
         token_count = reduce(lambda count, item: count + count_tokens(item)['n_tokens'],
                              map(lambda x: x["content"], filter(lambda x: "content" in x, messages)), 0)
-        response = make_openai_call(GPT_3_5_TURBO, messages, token_count, GRADING_FIELDS, GRADING_TEMPERATURE)
+        response = make_openai_call(GPT_3_5_TURBO, messages, token_count, None, TIPS_TEMPERATURE)
 
         if isinstance(response, str):
             response = re.sub(r"^[a-zA-Z0-9_]+\:\s*", "", response)
@@ -175,6 +187,7 @@ def get_writing_task_2_question():
     except Exception as e:
         return str(e)
 
+
 @app.route('/speaking_task_1', methods=['POST'])
 @jwt_required()
 def grade_speaking_task_1():
@@ -198,6 +211,7 @@ def grade_speaking_task_1():
         os.remove(sound_file_name)
         return str(e)
 
+
 @app.route('/speaking_task_1', methods=['GET'])
 @jwt_required()
 def get_speaking_task_1_question():
@@ -210,6 +224,7 @@ def get_speaking_task_1_question():
     except Exception as e:
         return str(e)
 
+
 @app.route('/speaking_task_2', methods=['POST'])
 @jwt_required()
 def grade_speaking_task_2():
@@ -233,6 +248,7 @@ def grade_speaking_task_2():
         os.remove(sound_file_name)
         return str(e)
 
+
 @app.route('/speaking_task_2', methods=['GET'])
 @jwt_required()
 def get_speaking_task_2_question():
@@ -245,5 +261,6 @@ def get_speaking_task_2_question():
     except Exception as e:
         return str(e)
 
+
 if __name__ == '__main__':
     app.run()
diff --git a/helper/api_messages.py b/helper/api_messages.py
index 3548a15..00a5ecc 100644
--- a/helper/api_messages.py
+++ b/helper/api_messages.py
@@ -349,30 +349,30 @@ def get_question_gen_messages(question_type: QuestionType):
 def get_question_tips(question: str, answer: str, correct_answer: str, context: str = None):
     messages = [
         {
-            "role": "system",
+            "role": "user",
             "content": "You are a IELTS exam program that analyzes incorrect answers to questions and gives tips to "
-                       "help students understand why it was a wrong answer and explains how their thought process "
-                       "should have been. The tip should refer to the context and question.",
+                       "help students understand why it was a wrong answer and gives helpful insight for the future. "
+                       "The tip should refer to the context and question.",
         }
     ]
 
     if not (context is None or context == ""):
         messages.append({
-            "role": "system",
+            "role": "user",
             "content": f"This is the context for the question: {context}",
         })
 
     messages.extend([
         {
-            "role": "system",
+            "role": "user",
             "content": f"This is the question: {question}",
         },
         {
-            "role": "system",
+            "role": "user",
             "content": f"This is the answer: {answer}",
         },
         {
-            "role": "system",
+            "role": "user",
             "content": f"This is the correct answer: {correct_answer}",
         }
     ])
diff --git a/helper/openai_interface.py b/helper/openai_interface.py
index edf8ee3..d5a4b57 100644
--- a/helper/openai_interface.py
+++ b/helper/openai_interface.py
@@ -53,7 +53,12 @@ def make_openai_call(model, messages, token_count, fields_to_check, temperature)
         frequency_penalty=float(FREQUENCY_PENALTY),
         messages=messages
     )
+
+    if fields_to_check is None:
+        return result["choices"][0]["message"]["content"]
+
     processed_response = process_response(result["choices"][0]["message"]["content"], fields_to_check[0])
+
     if check_fields(processed_response, fields_to_check) is False and try_count < TRY_LIMIT:
         try_count = try_count + 1
         return make_openai_call(model, messages, token_count, fields_to_check, temperature)