diff --git a/libs/langchain/langchain/indexes/_api.py b/libs/langchain/langchain/indexes/_api.py
index 6f2ebf9828..c471d07d9c 100644
--- a/libs/langchain/langchain/indexes/_api.py
+++ b/libs/langchain/langchain/indexes/_api.py
@@ -282,14 +282,14 @@ def index(
         # Filter out documents that already exist in the record store.
         uids = []
         docs_to_index = []
-        for doc, hashed_doc, doc_exists in zip(doc_batch, hashed_docs, exists_batch):
+        for hashed_doc, doc_exists in zip(hashed_docs, exists_batch):
             if doc_exists:
                 # Must be updated to refresh timestamp.
                 record_manager.update([hashed_doc.uid], time_at_least=index_start_dt)
                 num_skipped += 1
                 continue
             uids.append(hashed_doc.uid)
-            docs_to_index.append(doc)
+            docs_to_index.append(hashed_doc.to_document())
 
         # Be pessimistic and assume that all vector store write will fail.
         # First write to vector store
diff --git a/libs/langchain/tests/unit_tests/indexes/test_indexing.py b/libs/langchain/tests/unit_tests/indexes/test_indexing.py
index 7783f41f07..3567dd648c 100644
--- a/libs/langchain/tests/unit_tests/indexes/test_indexing.py
+++ b/libs/langchain/tests/unit_tests/indexes/test_indexing.py
@@ -472,3 +472,42 @@ def test_deduplication(
         "num_skipped": 0,
         "num_updated": 0,
     }
+
+
+def test_deduplication_v2(
+    record_manager: SQLRecordManager, vector_store: VectorStore
+) -> None:
+    """Check edge case when loader returns no new docs."""
+    docs = [
+        Document(
+            page_content="1",
+            metadata={"source": "1"},
+        ),
+        Document(
+            page_content="1",
+            metadata={"source": "1"},
+        ),
+        Document(
+            page_content="2",
+            metadata={"source": "2"},
+        ),
+        Document(
+            page_content="3",
+            metadata={"source": "3"},
+        ),
+    ]
+
+    # Should result in only a single document being added
+    assert index(docs, record_manager, vector_store, cleanup="full") == {
+        "num_added": 3,
+        "num_deleted": 0,
+        "num_skipped": 0,
+        "num_updated": 0,
+    }
+
+    # using in memory implementation here
+    assert isinstance(vector_store, InMemoryVectorStore)
+    contents = sorted(
+        [document.page_content for document in vector_store.store.values()]
+    )
+    assert contents == ["1", "2", "3"]