diff --git a/main.py b/main.py
index df65dcf..b1715ee 100644
--- a/main.py
+++ b/main.py
@@ -10,9 +10,11 @@ import time
 import json
 import uuid
 
-from utils import stream_message_template
+from utils import get_last_user_message, stream_message_template
+from schemas import OpenAIChatCompletionForm
 from config import MODEL_ID, MODEL_NAME
 
+from rag_pipeline import get_response
 
 app = FastAPI(docs_url="/docs", redoc_url=None)
 
@@ -58,48 +60,25 @@ async def get_models():
     }
 
 
-class OpenAIChatMessage(BaseModel):
-    role: str
-    content: str
-
-    model_config = ConfigDict(extra="allow")
-
-
-class OpenAIChatCompletionForm(BaseModel):
-    model: str
-    messages: List[OpenAIChatMessage]
-
-    model_config = ConfigDict(extra="allow")
-
-
-def get_response(user_message):
-    return f"rag response to: {user_message}"
-
-
 @app.post("/chat/completions")
 @app.post("/v1/chat/completions")
 async def generate_openai_chat_completion(form_data: OpenAIChatCompletionForm):
-
-    # get last user message (role == 'user') from the form_data
-    # last message might be role == 'assistant' or 'system' or 'user'
-    user_message = form_data.messages[-1].content
-
-    res = get_response(user_message)
-
-    finish_message = {
-        "id": f"rag-{str(uuid.uuid4())}",
-        "object": "chat.completion.chunk",
-        "created": int(time.time()),
-        "model": MODEL_ID,
-        "choices": [
-            {"index": 0, "delta": {}, "logprobs": None, "finish_reason": "stop"}
-        ],
-    }
+    user_message = get_last_user_message(form_data.messages)
+    res = get_response(user_message, messages=form_data.messages)
 
     def stream_content():
         message = stream_message_template(res)
-
         yield f"data: {json.dumps(message)}\n\n"
+
+        finish_message = {
+            "id": f"rag-{str(uuid.uuid4())}",
+            "object": "chat.completion.chunk",
+            "created": int(time.time()),
+            "model": MODEL_ID,
+            "choices": [
+                {"index": 0, "delta": {}, "logprobs": None, "finish_reason": "stop"}
+            ],
+        }
         yield f"data: {json.dumps(finish_message)}\n\n"
         yield f"data: [DONE]"
 
diff --git a/rag_pipeline.py b/rag_pipeline.py
new file mode 100644
index 0000000..1f74754
--- /dev/null
+++ b/rag_pipeline.py
@@ -0,0 +1,8 @@
+from typing import List
+from schemas import OpenAIChatMessage
+
+
+def get_response(user_message: str, messages: List[OpenAIChatMessage]):
+    print(messages)
+    print(user_message)
+    return f"rag response to: {user_message}"
diff --git a/schemas.py b/schemas.py
new file mode 100644
index 0000000..1c166ad
--- /dev/null
+++ b/schemas.py
@@ -0,0 +1,16 @@
+from typing import List
+from pydantic import BaseModel, ConfigDict
+
+
+class OpenAIChatMessage(BaseModel):
+    role: str
+    content: str
+
+    model_config = ConfigDict(extra="allow")
+
+
+class OpenAIChatCompletionForm(BaseModel):
+    model: str
+    messages: List[OpenAIChatMessage]
+
+    model_config = ConfigDict(extra="allow")
diff --git a/utils.py b/utils.py
index 3360a65..67af85c 100644
--- a/utils.py
+++ b/utils.py
@@ -1,5 +1,8 @@
 import uuid
 import time
+
+from typing import List
+from schemas import OpenAIChatMessage
 from config import MODEL_ID
 
 
@@ -18,3 +21,10 @@ def stream_message_template(message: str):
             }
         ],
     }
+
+
+def get_last_user_message(messages: List[OpenAIChatMessage]) -> str:
+    for message in reversed(messages):
+        if message.role == "user":
+            return message.content
+    return None