diff --git a/main.py b/main.py
index ddf4076..ee1911c 100644
--- a/main.py
+++ b/main.py
@@ -1,12 +1,14 @@
 from mentor_mingle.llm_handler import ChatHandler
-from mentor_mingle.persona.mentor import Mentor
 
 if __name__ == "__main__":
-    mentor = Mentor()
-    handler = ChatHandler(persona=mentor)
-
+    handler = ChatHandler()
     while True:
         user_prompt = input("\nUser: ")
         for res in handler.stream_chat(user_prompt):
             end_char = "\n" if "." in res else ""
             print(res, end=end_char, flush=True)
+
+        handler.last_chat = {
+            "query": user_prompt,
+            "assistant": "",
+        }
diff --git a/mentor_mingle/llm_handler.py b/mentor_mingle/llm_handler.py
index 1d010cc..44992f9 100644
--- a/mentor_mingle/llm_handler.py
+++ b/mentor_mingle/llm_handler.py
@@ -1,13 +1,15 @@
 import logging
 import os
+import redis
 from pathlib import Path
-from typing import Generator
+from typing import Generator, Optional, Union, Any
 
 import openai
 from dotenv import load_dotenv
 
 from .config import Config
-from .persona.base import BasePersona
+from .helpers.cache import Cache
+from .persona.mentor import Mentor
 from .utils import find_closest
 
 load_dotenv()
@@ -17,17 +19,44 @@
 class ChatHandler:
     """Handler for chat with GPT-3"""
 
+    _memory: Cache = None
+
     def __init__(
         self,
-        persona: BasePersona,
+        cache_client: Union[redis.Redis, Any] = None,
     ):
         """Initialize the chat handler"""
-        openai.api_key = os.getenv("OPENAI_KEY")
-        self.agent = persona
+        openai.api_key = os.environ.get("OPENAI_KEY")
+        self.agent = Mentor()
 
         # Load config
         self.model = Config.from_toml(Path(find_closest("config.toml"))).models.gpt3
 
+        # Initialize memory
+        if self._memory is None:
+            self._memory = Cache(cache_client)
+
+    @property
+    def memory(self) -> Cache:
+        """Get the memory"""
+        return self._memory
+
+    @property
+    def last_chat(self) -> Optional[dict]:
+        """Get the memory"""
+        return self._memory.get_map("last_chat")
+
+    @last_chat.setter
+    def last_chat(self, value: dict) -> None:
+        """Set the memory"""
+        self._memory.set_map(
+            "last_chat",
+            {
+                "query": value.get(b"query", ""),
+                "assistant": value.get(b"assistant", ""),
+            },
+        )
+
     def stream_chat(self, user_prompt: str) -> Generator[str, None, None]:
         """
         Stream a chat with GPT-3
@@ -38,15 +67,27 @@ def stream_chat(self, user_prompt: str) -> Generator[str, None, None]:
         Returns:
             None
         """
-        completion = openai.ChatCompletion.create(
-            model=self.model.name,
-            messages=[
-                {"role": "system", "content": self.agent.persona},
+        intro_session = [
+            {"role": "system", "content": self.agent.persona},
+            {
+                "role": "user",
+                "content": f"User: {user_prompt}",
+            },
+            {
+                "role": "system",
+                "content": self.agent.answer_format,
+            },
+        ]
+        if self.last_chat:
+            intro_session += [
                 {
                     "role": "user",
-                    "content": f"User: {user_prompt}" f"\n{self.agent.answer_format}",
-                },
-            ],
+                    "content": f"{self.last_chat.get(b'query', '')}",
+                }
+            ]
+        completion = openai.ChatCompletion.create(
+            model=self.model.name,
+            messages=intro_session,
             **self.model.config.model_dump(),
         )
         for chunk in completion:
diff --git a/tests/conftest.py b/tests/conftest.py
new file mode 100644
index 0000000..a6e569b
--- /dev/null
+++ b/tests/conftest.py
@@ -0,0 +1,12 @@
+
+import pytest
+import fakeredis
+
+@pytest.fixture(scope="session")
+def fake_redis() -> fakeredis.FakeStrictRedis:
+    """
+    Create a fake Redis client.
+
+    Returns: FakeRedis client
+    """
+    return fakeredis.FakeStrictRedis()
\ No newline at end of file
diff --git a/tests/helpers/test_cache.py b/tests/helpers/test_cache.py
index 6ae21ea..3c7f343 100644
--- a/tests/helpers/test_cache.py
+++ b/tests/helpers/test_cache.py
@@ -1,5 +1,3 @@
-import fakeredis
-import pytest
 
 from mentor_mingle.helpers.cache import Cache
 
@@ -9,15 +7,6 @@ class TestCache:
     Test the Cache class.
     """
 
-    @pytest.fixture
-    def fake_redis(self) -> fakeredis.FakeStrictRedis:
-        """
-        Create a fake Redis client.
-
-        Returns: FakeRedis client
-        """
-        return fakeredis.FakeStrictRedis()
-
     def test_get(self, fake_redis):
         """
         Test the get method of the Cache class.
diff --git a/tests/test_llm_handler.py b/tests/test_llm_handler.py
index 1077b5a..f5c5439 100644
--- a/tests/test_llm_handler.py
+++ b/tests/test_llm_handler.py
@@ -13,11 +13,11 @@ class TestChatHandler:
     Test the ChatHandler class.
     """
 
-    def test_chat_handler_init(self):
+    def test_chat_handler_init(self, fake_redis):
         """
         Test the stream_chat method of the ChatHandler class.
         """
-        handler = ChatHandler(Mentor())
+        handler = ChatHandler(cache_client=fake_redis)
         assert isinstance(handler.model, Gpt)
         assert isinstance(handler.agent, Mentor)
         assert isinstance(handler.model.config, CFGGpt)
@@ -40,7 +40,7 @@ def mock_response_generator(self, **kwargs) -> Generator[OpenAIObject, None, Non
             mock_obj.choices = [choice_mock]
             yield mock_obj
 
-    def test_stream_chat(self, mocker: MagicMock):
+    def test_stream_chat(self, mocker: MagicMock, fake_redis):
         """
         Test the stream_chat method of the ChatHandler class.
 
@@ -51,7 +51,7 @@ def test_stream_chat(self, mocker: MagicMock):
             None
         """
         # Create a mock instance of your class
-        llm = ChatHandler(Mentor())
+        llm = ChatHandler(cache_client=fake_redis)
 
         # Patch the API call to return mock_response
         mocker.patch.object(openai.ChatCompletion, "create", side_effect=self.mock_response_generator)