diff --git a/common/gsql/supportai/retrievers/GraphRAG_Community_Vector_Search.gsql b/common/gsql/supportai/retrievers/GraphRAG_Community_Vector_Search.gsql
index 29594d8..46a433f 100644
--- a/common/gsql/supportai/retrievers/GraphRAG_Community_Vector_Search.gsql
+++ b/common/gsql/supportai/retrievers/GraphRAG_Community_Vector_Search.gsql
@@ -37,6 +37,10 @@ CREATE OR REPLACE DISTRIBUTED QUERY GraphRAG_Community_Vector_Search(LIST<FLOAT>
               ACCUM s.@children += c
               POST-ACCUM @@verbose_info += ("community_level_"+to_string(i-1) -> s.@children);
       END;
+      start_chunks = vectorSearch({DocumentChunk.embedding}, query_vector, top_k);
+      extra_selected_comms = SELECT m FROM start_chunks:dc -(CONTAINS_ENTITY>)- Entity:v -(RESOLVES_TO>)- ResolvedEntity:r -(IN_COMMUNITY>)- Community:m;
+      selected_comms = selected_comms UNION extra_selected_comms;
+
       IF with_doc THEN
           related_chunks = SELECT c FROM Content:c -(<HAS_CONTENT)- Document:d -(HAS_CHILD>)- DocumentChunk:dc -(CONTAINS_ENTITY>)- Entity:v -(RESOLVES_TO>)- ResolvedEntity:r -(IN_COMMUNITY>)- selected_comms:m
               ACCUM m.@context += c.text, m.@children += d
diff --git a/graphrag/app/routers/supportai.py b/graphrag/app/routers/supportai.py
index 4bba7dd..5c377eb 100644
--- a/graphrag/app/routers/supportai.py
+++ b/graphrag/app/routers/supportai.py
@@ -238,6 +238,8 @@ def search(
             query.method_params["with_doc"],
             query.method_params["verbose"],
         )
+    else:
+        raise Exception(f"Method {query.method} not implemented")
     return res
 
 
diff --git a/graphrag/app/supportai/retrievers/CommunityRetriever.py b/graphrag/app/supportai/retrievers/CommunityRetriever.py
index fab69d1..549c58d 100644
--- a/graphrag/app/supportai/retrievers/CommunityRetriever.py
+++ b/graphrag/app/supportai/retrievers/CommunityRetriever.py
@@ -62,6 +62,22 @@ def search(self, question, community_level: int, top_k: int = 5, similarity_thre
                 },
                 usePost=True
             )
+
+            # Include similarity search results
+            if with_chunk or with_doc:
+                start_set = self._generate_start_set(questions, ["DocumentChunk"], top_k, similarity_threshold)
+
+                self._check_query_install("Content_Similarity_Search")
+                resp = self.conn.runInstalledQuery(
+                    "Content_Similarity_Search",
+                    params = {
+                        "json_list_vts": str(start_set),
+                        "v_type": "DocumentChunk",
+                        "verbose": verbose,
+                    },
+                    usePost=True
+                )
+                res[0]["final_retrieval"]["Similarity_Context"] = [resp[0]["final_retrieval"][x] for x in resp[0]["final_retrieval"]]
         else:
             query_vector = self._generate_embedding(question)