diff --git a/apps/base/customer1/openclaw/a100-vllm.yaml b/apps/base/customer1/openclaw/a100-vllm.yaml
index f34df7d..5e565ef 100644
--- a/apps/base/customer1/openclaw/a100-vllm.yaml
+++ b/apps/base/customer1/openclaw/a100-vllm.yaml
@@ -39,12 +39,12 @@ spec:
             - --tokenizer-mode=hf 
             - --gpu-memory-utilization=0.95
             - --kv-cache-dtype=fp8
-            - --max-model-len=49152 
+            - --max-model-len=auto 
             - --enable-auto-tool-choice 
             - --trust-remote-code
             - --dtype=auto
             - --enable-prefix-caching
-            - --tool-call-parser=hermes
+            - --tool-call-parser=qwen3_coder
             - --reasoning-parser=qwen3
             
             - --disable-custom-all-reduce
diff --git a/apps/base/customer1/openclaw/configmap.yaml b/apps/base/customer1/openclaw/configmap.yaml
index 6727d13..eed4ca8 100644
--- a/apps/base/customer1/openclaw/configmap.yaml
+++ b/apps/base/customer1/openclaw/configmap.yaml
@@ -71,9 +71,16 @@ data:
           }
         ]
       },
-      "cron": { "enabled": true } 
+      "cron": { "enabled": true },
+      "requestTimeoutSeconds": 180,
+      "promptCaching": true,
+      "context": {
+        "injectAllMarkdown": false,
+        "curatedFiles": ["SOUL.md", "MEMORY.md", "IDENTITY.md", "USER.md"],
+        "useMemorySearch": true
+      }
     }
   AGENTS.md: |
-    # OpenClaw Assistant
+    # Sirius Claw
 
-    You are a helpful AI assistant running in Kubernetes.
+    Professional, precise, dark humor. Running in Lance's GKE cluster.
diff --git a/apps/base/customer1/openclaw/keda-vllm.yaml b/apps/base/customer1/openclaw/keda-vllm.yaml
new file mode 100644
index 0000000..c26cf69
--- /dev/null
+++ b/apps/base/customer1/openclaw/keda-vllm.yaml
@@ -0,0 +1,21 @@
+apiVersion: keda.sh/v1alpha1
+kind: ScaledObject
+metadata:
+  name: openclaw-brain-a100-scaling
+  namespace: customer1
+spec:
+  scaleTargetRef:
+    name: openclaw-brain-vllm
+    kind: Deployment
+  service:
+    name: openclaw-brain-service
+    port: 8000
+  minReplicaCount: 0
+  maxReplicaCount: 1
+  cooldownPeriod: 900  # 15 minutes idle before scale-down
+  triggers:
+    - type: http
+      metadata:
+        path: "/v1/models"
+        host: "openclaw-brain-service.customer1.svc.cluster.local"
+        requestCount: "1"  # Scale up on 1+ request (wake script handles cold start)
diff --git a/apps/base/customer1/openclaw/kustomization.yaml b/apps/base/customer1/openclaw/kustomization.yaml
index 546fdd1..5c7138d 100644
--- a/apps/base/customer1/openclaw/kustomization.yaml
+++ b/apps/base/customer1/openclaw/kustomization.yaml
@@ -2,7 +2,7 @@ apiVersion: kustomize.config.k8s.io/v1beta1
 kind: Kustomization
 resources:
  # - vllm-l4.yaml
-  #- keda-a100.yaml
+  - keda-vllm.yaml
   - pvc.yaml
   - configmap.yaml
   - deployment.yaml