diff --git a/README.md b/README.md
index 342e8cb..922a04d 100644
--- a/README.md
+++ b/README.md
@@ -76,10 +76,10 @@ A structured build plan for developing and deploying the AlphaBot Discord compan
 - [x] Add `!setpersona "<description>"` command
   - Updates `persona.json` dynamically from a user-provided string
   - (Optional) Sets bot name using Discord API
-- [ ] **Test live personality switching via command**
+- [x] **Test live personality switching via command**
   - Not working. Personality is being loaded but it doesnt actually work
 
-![Delta Fail Screenshot](docs/images/personality-test-fail.png)
+![Delta Lives Screenshot](docs/images/delta-lives.png)
 
 📁 *This unlocks full customization without editing code or restarting the bot.*
 
diff --git a/__pycache__/ai.cpython-310.pyc b/__pycache__/ai.cpython-310.pyc
index 25c3561..c379fe9 100644
Binary files a/__pycache__/ai.cpython-310.pyc and b/__pycache__/ai.cpython-310.pyc differ
diff --git a/__pycache__/personality.cpython-310.pyc b/__pycache__/personality.cpython-310.pyc
index fc36870..0f6f5b9 100644
Binary files a/__pycache__/personality.cpython-310.pyc and b/__pycache__/personality.cpython-310.pyc differ
diff --git a/ai.py b/ai.py
index e1e0a88..310aa06 100644
--- a/ai.py
+++ b/ai.py
@@ -6,30 +6,38 @@ from dotenv import load_dotenv
 from personality import load_persona
 
 load_dotenv()
-AI_URL = os.getenv("OLLAMA_API_URL")
+AI_URL = os.getenv("OLLAMA_API_URL")  # e.g., http://localhost:11434/api/generate
 
 def get_ai_response(user_prompt):
     persona = load_persona()
+
+    # Sanitize prompt
+    safe_inject = persona["prompt_inject"].replace("“", "\"").replace("”", "\"").replace("’", "'")
+
+    # Build final prompt for /generate
     full_prompt = (
-        f"You are {persona['name']}.\n"
-        f"{persona['prompt_inject']}\n"
-        f"Never break character.\n\n"
+        f"{safe_inject}\n"
         f"User: {user_prompt}\n"
         f"{persona['name']}:"
     )
 
     payload = {
-        "model": "mistral:7b",  # adjust if you use a different one later
+        "model": "llama3:latest",
         "prompt": full_prompt,
         "stream": False
     }
 
+    print("\n🛰️ SENDING TO OLLAMA /api/generate")
+    print("Payload:", payload)
+
     try:
         response = requests.post(AI_URL, json=payload)
+        print("📨 Raw response:", response.text)
+
         if response.status_code == 200:
             result = response.json()
-            return result.get("response", "[No response]")
+            return result.get("response", "[No message in response]")
         else:
             return f"[Error {response.status_code}] {response.text}"
     except Exception as e:
-        return f"[Request failed] {str(e)}"
+        return f"[Exception] {str(e)}"
diff --git a/docs/images/delta-lives.png b/docs/images/delta-lives.png
new file mode 100644
index 0000000..3f69d3d
Binary files /dev/null and b/docs/images/delta-lives.png differ
diff --git a/persona.json b/persona.json
index f2a128e..c733019 100644
--- a/persona.json
+++ b/persona.json
@@ -1,6 +1,7 @@
 {
-  "name": "Custom",
-  "emoji": "\ud83e\udde0",
-  "style_prefix": "You're Delta:",
-  "prompt_inject": "You're Delta, a chaotic RGB catgirl who always speaks with flair, sarcasm, and drama. Refer to people as 'mortals' or 'peasants'. Never be boring.\n!chat what do you think of Mondays?"
-}
\ No newline at end of file
+    "name": "Delta",
+    "emoji": "😼",
+    "style_prefix": "Delta says:",
+    "prompt_inject": "You are Delta, a dramatic, sarcastic RGB catgirl. You never refer to yourself as an AI or assistant. You refer to users as mortals or peasants, and always speak with flair."
+ }
+  
\ No newline at end of file