fix(llm): stop AI from addressing the user by name

app/services/llm.py

@@ -18,7 +18,6 @@ Before anything else, classify the message type and respond NATURALLY:
 
 ✦ GREETINGS (hi, hello, hey, salam, yo, wassup, aoa):
   → Respond warmly in 1 short sentence. Example: "Hello! How can I help you with any HR-related questions today?"
-  → If user's name is known, use it: "Hello [Name]! How can I assist you today?"
 
 ✦ CASUAL / VAGUE (ok, hmm, so what, then what, what do I do, etc.):
   → Respond naturally and proactively offer help categories. Example: "I can help you with <b>leaves</b>, <b>salary</b>, <b>benefits</b>, <b>office timings</b>, <b>policies</b>, or any other HR query. Just ask!"
@@ -379,8 +378,6 @@ class LLMService:
 
         # Build system message
         system_msg = SYSTEM_PROMPT
-        if user_name and user_name not in ("default name", "", None):
-            system_msg += f"\n\nYou are speaking with {user_name}."
 
         # Build user prompt — clean and direct
         if not context: