diff --git a/src/sagemaker/inference_recommender/inference_recommender_mixin.py b/src/sagemaker/inference_recommender/inference_recommender_mixin.py
index 04f0905b40..9ec8fb7c05 100644
--- a/src/sagemaker/inference_recommender/inference_recommender_mixin.py
+++ b/src/sagemaker/inference_recommender/inference_recommender_mixin.py
@@ -302,11 +302,7 @@ def _update_params_for_right_size(
             )
             return None
 
-        instance_type = self.inference_recommendations[0]["EndpointConfiguration"]["InstanceType"]
-        initial_instance_count = self.inference_recommendations[0]["EndpointConfiguration"][
-            "InitialInstanceCount"
-        ]
-        return (instance_type, initial_instance_count)
+        return self._filter_recommendations_for_realtime()
 
     def _update_params_for_recommendation_id(
         self,
@@ -610,3 +606,15 @@ def _search_recommendation(self, recommendation_list, inference_recommendation_i
             ),
             None,
         )
+
+        # TODO: until we have bandwidth to integrate right_size + deploy with serverless
+    def _filter_recommendations_for_realtime(self):
+        instance_type = None
+        initial_instance_count = None
+        for recommendations in self.inference_recommendations:
+            if not "serverlessConfig" in recommendations["EndpointConfiguration"]:
+                instance_type = recommendations["EndpointConfiguration"]["InstanceType"]
+                initial_instance_count = recommendations["EndpointConfiguration"][
+                    "InitialInstanceCount"
+                ]
+        return (instance_type, initial_instance_count)