diff --git a/core/faiss_index.bin b/core/faiss_index.bin
new file mode 100644
index 0000000..d3f0f6b
Binary files /dev/null and b/core/faiss_index.bin differ
diff --git a/core/graphAgent.py b/core/graphAgent.py
index c6605e5..09278c7 100644
--- a/core/graphAgent.py
+++ b/core/graphAgent.py
@@ -31,6 +31,7 @@ def __init__(self):
         self.workflow.add_node("use_calendar_tool", calendar_tool_decider)
         self.workflow.add_node("calendar_decider", calendar_decision_agent)
         self.workflow.add_node("other_agent", other_agent)
+        
 
         self.workflow.add_edge(START, "jarvis_agent")
         self.workflow.add_edge("perplexity_agent", "tools")
diff --git a/core/graphstate.py b/core/graphstate.py
index dd1ea16..2058c62 100644
--- a/core/graphstate.py
+++ b/core/graphstate.py
@@ -10,5 +10,4 @@ class GraphState(TypedDict):
     data: dict
     tool_decision: Literal["use_tool", "generate"]
     agent_decision: Literal["perplexity", "calendar", "other"]
-    calendar_decision: Literal["use_calendar_tool", "return_to_jarvis"]
-    
+    calendar_decision: Literal["use_calendar_tool", "return_to_jarvis"]
\ No newline at end of file
diff --git a/core/id_to_metadata.json b/core/id_to_metadata.json
new file mode 100644
index 0000000..0228f8e
--- /dev/null
+++ b/core/id_to_metadata.json
@@ -0,0 +1 @@
+{"210453397504": {"user_id": "1", "text": "Samtale fra 2024-11-07:\n{\"chat_summary\":\"The human greeted the AI with 'hi' twice, and the AI responded with 'success' both times.\"}"}, "210453397505": {"user_id": "1", "text": "Samtale fra 2024-11-07:\n{\"chat_summary\":\"The human bought a new red Audi A7 car yesterday.\"}"}}
\ No newline at end of file
diff --git a/core/main.py b/core/main.py
index 8f39685..80254ac 100644
--- a/core/main.py
+++ b/core/main.py
@@ -12,6 +12,7 @@
 import logging
 log = logging.getLogger('werkzeug')
 log.setLevel(logging.ERROR)
+from collections import defaultdict
 
 #
 #   Setup
@@ -31,6 +32,9 @@
 # Agent instantiation
 jarvis = Graph() # API key is configured in agent.py
 
+# Initialize active_chats with the correct format
+active_chats = defaultdict(lambda: {"chat_history": []})
+
 #
 #
 #   HTTP(S) routes below
@@ -67,27 +71,71 @@ def summarize_store():
 # Base event that's fired when a user connects
 @socketio.on('connect') 
 def connect(data):
+    session_id = request.sid
     emit("You're connected to Jarvis streaming server...")
     print('UI connected to backend')
+    print(f'Session ID: {session_id}')
 
 # Base event that's fired when user gracefully disconnects
 @socketio.on('disconnect')
 def disconnect():
+    session_id = request.sid
+    if session_id in active_chats:
+        # Get the chat history before deleting it
+        chat_history = active_chats[session_id]
+        
+        try:
+            # Call summarize_chat directly instead of the route handler
+            summary = summarize_chat(chat_history)
+            # Then call embed_and_store directly
+            embed_and_store(summary, "1")  # Using dummy user_id "1"
+            print(f'Chat history summarized and stored')
+        except Exception as e:
+            print(f'Error summarizing chat history: {e}')
+        
+        # Clean up the chat history
+        del active_chats[session_id]
+    
     print('UI disconnected')
+    print(f'Session ID: {session_id}')
 
 # Custom event. Fired when the user sends a prompt.
 @socketio.on('user_prompt')
 def handle_prompt(data):
     try:
-        conversation_id = data['conversation_id'] # grabs the conversation ID
+        session_id = request.sid
+        conversation_id = data['conversation_id']
+        
+        # Create new chat entry with human message
+        chat_entry = {
+            "human_message": data['prompt'],
+            "ai_message": ""  # Will be filled when AI responds
+        }
+        
         socketio.emit("start_message")
-        asyncio.run(jarvis.run(data['prompt'], socketio), debug=True) # prompts Jarvis and hands off emitting to the graphAgent.
+        
+        # Run the AI response
+        async def run_and_store():
+            response = await jarvis.run(data['prompt'], socketio)
+            # Update the AI response in the chat entry
+            chat_entry["ai_message"] = response
+            # Add completed chat entry to history
+            active_chats[session_id]["chat_history"].append(chat_entry)
+            
+        asyncio.run(run_and_store(), debug=True)
         
         return jsonify({"status": "success"})
     except Exception as e:
         print(f'Something very bad happened: {e}')
         return jsonify({"status": "error"})
 
+@socketio.on('get_chat_history')
+def get_chat_history():
+    session_id = request.sid
+    if session_id in active_chats:
+        return active_chats[session_id]
+    return {"chat_history": []}
+
 if __name__ == '__main__':
     socketio.run(app, debug=True, host='0.0.0.0', port=PORT, allow_unsafe_werkzeug=True)
 
diff --git a/core/noder.py b/core/noder.py
index 5cd04ef..207edf7 100644
--- a/core/noder.py
+++ b/core/noder.py
@@ -12,7 +12,7 @@ def jarvis_agent(state: GraphState):
         template= """
         Your job is to determine if you need tools to answer the
         users question and answer with only the name of the option
-        chosen.
+        chosen. You have access to chat history using tools, thus also some personal data can be retrieved.
 
         Here are previous messages:
         
@@ -53,7 +53,7 @@ def tool_agent_decider(state: GraphState):
 
         Your options for agents are the following:
         - "perplexity": This agent has access to tools that use the perplexity API and tools 
-                        for doing a RAG-search. These tools are the following: {perplexity_tools}
+                        for doing a chat history search. These tools are the following: {perplexity_tools}
         - "calendar":   This agent has access to calendar tools
                          These tools are the following: {calendar_tools}
         - "other":      Other tools available: {other_tools}
@@ -113,7 +113,7 @@ def perplexity_agent(state: GraphState):
     prompt = PromptTemplate(
         template= """
         Your job is to create tool_calls to tools using the perplexity API or 
-        to tools that do a RAG-search. The tool or tools you decide
+        to tools that do a RAG-search on the chat history. The tool or tools you decide
         to call should help answer the users question.
 
         Here are previous messages:
@@ -202,6 +202,7 @@ def other_agent(state: GraphState):
         Your job is to create tool_calls to tools.
         The tool or tools you decide
         to call should help answer the users question.
+        You can also use the chat history search tool to help answer the users question.
 
         Here are previous messages:
         
diff --git a/core/static/socketEvents.js b/core/static/socketEvents.js
index d0d9da2..aa9c4e5 100644
--- a/core/static/socketEvents.js
+++ b/core/static/socketEvents.js
@@ -3,8 +3,16 @@
 // TODO: add this port to .env later
 var socket = io("ws://localhost:3000"); // websocket querying port 3001, where Flask is running.
 
-socket.on("connect", () => {
-  socket.emit("message", { data: "I'm connected!" });
+socket.on('connect', () => {
+    socket.emit('message', {data: 'I\'m connected!'});
+    
+    socket.emit('get_chat_history', (history) => {
+        // Restore chat history to UI
+        history.chat_history.forEach(entry => {
+            addUserMessage(entry.human_message);
+            addMessage(entry.ai_message);
+        });
+    });
 });
 
 let tokenCounter = 0;
diff --git a/core/tools/rag_search.py b/core/tools/rag_search.py
new file mode 100644
index 0000000..7e6dfc4
--- /dev/null
+++ b/core/tools/rag_search.py
@@ -0,0 +1,24 @@
+from rag import similarity_search
+from langchain_core.tools import tool
+from langchain_core.tools.structured import StructuredTool
+from typing import List, Tuple
+
+
+@tool 
+def rag_search(query: str)-> List[Tuple[str, float]]:
+    """
+    Use this tool to get relevant conversation history
+
+    Args:
+        query (str): The query to search for 
+
+    Returns:
+        List[Tuple[str, float]]: Relevant chat history 
+    """
+    result = similarity_search(query, "1")
+
+    return result
+
+
+def get_tool() -> StructuredTool:
+    return rag_search
\ No newline at end of file
diff --git a/core/tools/tools.py b/core/tools/tools.py
index 84d35bb..0d78d62 100644
--- a/core/tools/tools.py
+++ b/core/tools/tools.py
@@ -10,6 +10,7 @@
 import tools.create_time_to_iso as create_time_to_iso_format
 import tools.current_time_iso as current_time_iso_format
 import tools.add_time as add_time
+import tools.rag_search as rag_search
 
 
 def get_tools() -> list[StructuredTool]:
@@ -24,6 +25,7 @@ def get_tools() -> list[StructuredTool]:
     tools.append(read_calendar_event.get_tool())
     tools.append(create_time_to_iso_format.get_tool())
     tools.append(current_time_iso_format.get_tool())
+    tools.append(rag_search.get_tool())
 
     return tools
 
@@ -31,6 +33,7 @@ def get_perplexity_based_tools() -> list[StructuredTool]:
     tools = []
     tools.append(weather.get_tool())
     tools.append(web_search.get_tool())
+    tools.append(rag_search.get_tool())
 
     return tools
 
@@ -49,5 +52,6 @@ def get_other_tools() -> list[StructuredTool]:
     tools.append(find_files.get_tool())
     tools.append(read_file.get_tool())
     tools.append(read_pdf.get_tool())
+    tools.append(rag_search.get_tool())
 
     return tools
\ No newline at end of file
diff --git a/core/tools/weather.py b/core/tools/weather.py
index bea4d8c..8c3651a 100644
--- a/core/tools/weather.py
+++ b/core/tools/weather.py
@@ -34,7 +34,7 @@ def weather_forecast(user_query: str) -> str:
 
     # chat completion without streaming
     response = client.chat.completions.create(
-        model="llama-3.1-sonar-large-128k-online",
+        model="llama-3.1-sonar-huge-128k-online",
         messages=messages,
         temperature=0.0,
     )
diff --git a/core/user_to_vector_ids.json b/core/user_to_vector_ids.json
new file mode 100644
index 0000000..b64e184
--- /dev/null
+++ b/core/user_to_vector_ids.json
@@ -0,0 +1 @@
+{"1": [210453397504, 210453397505]}
\ No newline at end of file