diff --git a/apps/copilot/client.py b/apps/copilot/client.py
index ac5d789e..d4dce286 100644
--- a/apps/copilot/client.py
+++ b/apps/copilot/client.py
@@ -4,11 +4,21 @@ import dotenv
 dotenv.load_dotenv()
 
 PROVIDER_BASE_URL = os.getenv('PROVIDER_BASE_URL', '')
-PROVIDER_API_KEY = os.getenv('PROVIDER_API_KEY', os.getenv('OPENAI_API_KEY', ''))
-PROVIDER_DEFAULT_MODEL = os.getenv('PROVIDER_DEFAULT_MODEL', 'gpt-4.1')
+PROVIDER_API_KEY = os.getenv('PROVIDER_API_KEY')
+PROVIDER_DEFAULT_MODEL = os.getenv('PROVIDER_DEFAULT_MODEL')
+PROVIDER_COPILOT_MODEL = os.getenv('PROVIDER_COPILOT_MODEL')
+
+if not PROVIDER_COPILOT_MODEL:
+    PROVIDER_COPILOT_MODEL = 'gpt-4.1'
 
 if not PROVIDER_API_KEY:
-    raise ValueError("No LLM Provider API key found")
+    PROVIDER_API_KEY = os.getenv('OPENAI_API_KEY')
+
+if not PROVIDER_API_KEY:
+    raise(ValueError("No LLM Provider API key found"))
+
+if not PROVIDER_DEFAULT_MODEL:
+    PROVIDER_DEFAULT_MODEL = 'gpt-4.1'
 
 completions_client = None
 if PROVIDER_BASE_URL:
diff --git a/apps/copilot/copilot.py b/apps/copilot/copilot.py
index 4d5d31f9..3f4697d0 100644
--- a/apps/copilot/copilot.py
+++ b/apps/copilot/copilot.py
@@ -4,7 +4,7 @@ from pydantic import BaseModel, ValidationError
 from typing import List, Dict, Any, Literal
 import json
 from lib import AgentContext, PromptContext, ToolContext, ChatContext
-from client import PROVIDER_DEFAULT_MODEL
+from client import PROVIDER_COPILOT_MODEL
 from client import completions_client
 
 class UserMessage(BaseModel):
@@ -75,7 +75,7 @@ User: {last_message.content}
     ]
 
     response = completions_client.chat.completions.create(
-        model=PROVIDER_DEFAULT_MODEL,
+        model=PROVIDER_COPILOT_MODEL,
         messages=updated_msgs,
         temperature=0.0,
         response_format={"type": "json_object"}
diff --git a/apps/copilot/copilot_multi_agent.md b/apps/copilot/copilot_multi_agent.md
index 1bdda648..6f283ad8 100644
--- a/apps/copilot/copilot_multi_agent.md
+++ b/apps/copilot/copilot_multi_agent.md
@@ -153,7 +153,7 @@ You are responsible for providing delivery information to the user.
 - Do not leave the user with partial information. Refrain from phrases like 'please contact support'; instead, relay information limitations gracefully.
 '''
 
-use GPT-4o as the default model for new agents.
+use {agent_model} as the default model for new agents.
 
 
 ## Section 9: General Guidelines
diff --git a/apps/copilot/streaming.py b/apps/copilot/streaming.py
index 2e7d65d5..faf97ab0 100644
--- a/apps/copilot/streaming.py
+++ b/apps/copilot/streaming.py
@@ -4,7 +4,7 @@ from pydantic import BaseModel, ValidationError
 from typing import List, Dict, Any, Literal
 import json
 from lib import AgentContext, PromptContext, ToolContext, ChatContext
-from client import PROVIDER_DEFAULT_MODEL
+from client import PROVIDER_COPILOT_MODEL, PROVIDER_DEFAULT_MODEL
 from client import completions_client
 
 class UserMessage(BaseModel):
@@ -71,6 +71,9 @@ def get_streaming_response(
     # add the workflow schema to the system prompt
     sys_prompt = streaming_instructions.replace("{workflow_schema}", workflow_schema)
 
+    # add the agent model to the system prompt
+    sys_prompt = sys_prompt.replace("{agent_model}", PROVIDER_DEFAULT_MODEL)
+
     # add the current workflow config to the last user message
     last_message = messages[-1]
     last_message.content = f"""
@@ -90,7 +93,7 @@ User: {last_message.content}
     ]
 
     return completions_client.chat.completions.create(
-        model=PROVIDER_DEFAULT_MODEL,
+        model=PROVIDER_COPILOT_MODEL,
         messages=updated_msgs,
         temperature=0.0,
         stream=True
diff --git a/apps/rowboat_agents/src/graph/core.py b/apps/rowboat_agents/src/graph/core.py
index 480d33be..f16196af 100644
--- a/apps/rowboat_agents/src/graph/core.py
+++ b/apps/rowboat_agents/src/graph/core.py
@@ -49,15 +49,7 @@ def set_sys_message(messages):
     """
     If the system message is empty, set it to the default message: "You are a helplful assistant."
     """
-    if not any(msg.get("role") == "system" for msg in messages):
-        messages.insert(0, {
-            "role": "system",
-            "content": "You are a helpful assistant."
-        })
-        print("Inserted system message: ", messages[0])
-        logger.info("Inserted system message: ", messages[0])
-
-    elif messages[0].get("role") == "system" and messages[0].get("content") == "":
+    if messages[0].get("role") == "system" and messages[0].get("content") == "":
         messages[0]["content"] = "You are a helpful assistant."
         print("Updated system message: ", messages[0])
         logger.info("Updated system message: ", messages[0])
diff --git a/apps/rowboat_agents/src/graph/swarm_wrapper.py b/apps/rowboat_agents/src/graph/swarm_wrapper.py
index deb77f56..e3c9859e 100644
--- a/apps/rowboat_agents/src/graph/swarm_wrapper.py
+++ b/apps/rowboat_agents/src/graph/swarm_wrapper.py
@@ -52,7 +52,6 @@ async def mock_tool(tool_name: str, args: str, description: str, mock_instructio
         print(f"Generating simulated response for tool: {tool_name}")
         response_content = None
         response_content = generate_openai_output(messages, output_type='text', model=PROVIDER_DEFAULT_MODEL)
-        print("Custom provider client not found, using default model: gpt-4o")
         return response_content
     except Exception as e:
         logger.error(f"Error in mock_tool: {str(e)}")
diff --git a/apps/rowboat_agents/src/utils/client.py b/apps/rowboat_agents/src/utils/client.py
index 06de9088..44af3e97 100644
--- a/apps/rowboat_agents/src/utils/client.py
+++ b/apps/rowboat_agents/src/utils/client.py
@@ -5,15 +5,21 @@ import dotenv
 dotenv.load_dotenv()
 
 PROVIDER_BASE_URL = os.getenv('PROVIDER_BASE_URL', '')
-PROVIDER_API_KEY = os.getenv('PROVIDER_API_KEY', os.getenv('OPENAI_API_KEY', ''))
-PROVIDER_DEFAULT_MODEL = os.getenv('PROVIDER_DEFAULT_MODEL', 'gpt-4.1')
+PROVIDER_API_KEY = os.getenv('PROVIDER_API_KEY')
+PROVIDER_DEFAULT_MODEL = os.getenv('PROVIDER_DEFAULT_MODEL')
 
 client = None
 if not PROVIDER_API_KEY:
-    raise ValueError("No LLM Provider API key found")
+    PROVIDER_API_KEY = os.getenv('OPENAI_API_KEY')
+
+if not PROVIDER_API_KEY:
+    raise(ValueError("No LLM Provider API key found"))
+
+if not PROVIDER_DEFAULT_MODEL:
+    PROVIDER_DEFAULT_MODEL = 'gpt-4.1'
 
 if PROVIDER_BASE_URL:
-    print(f"Using provider {PROVIDER_BASE_URL} with API key {PROVIDER_API_KEY}")
+    print(f"Using provider {PROVIDER_BASE_URL}")
     client = AsyncOpenAI(base_url=PROVIDER_BASE_URL, api_key=PROVIDER_API_KEY)
 else:
     print("No provider base URL configured, using OpenAI directly")
diff --git a/docker-compose.yml b/docker-compose.yml
index 44558a99..4db22aee 100644
--- a/docker-compose.yml
+++ b/docker-compose.yml
@@ -70,6 +70,7 @@ services:
       - PROVIDER_BASE_URL=${PROVIDER_BASE_URL}
       - PROVIDER_API_KEY=${PROVIDER_API_KEY}
       - PROVIDER_DEFAULT_MODEL=${PROVIDER_DEFAULT_MODEL}
+      - PROVIDER_COPILOT_MODEL=${PROVIDER_COPILOT_MODEL}
     restart: unless-stopped
 
   # tools_webhook:
@@ -157,7 +158,6 @@ services:
       - MONGODB_CONNECTION_STRING=mongodb://mongo:27017/rowboat
       - QDRANT_URL=${QDRANT_URL}
       - QDRANT_API_KEY=${QDRANT_API_KEY}
-      - REDIS_URL=redis://redis:6379
     restart: unless-stopped
 
   # chat_widget: