Noticed some wording stuff

docs/usage.md

@@ -306,7 +306,7 @@ Paperless-ngx includes several features that use AI to enhance the document mana
     so consider the privacy implications of using these features, especially if using a remote
     model or API provider instead of the default local model.
 
-The AI features work by creating an embedding of the text content and metadata of documents, which is then used for various tasks such as similarity search and question answering. This uses the FAISS vector store.
+The AI features work by creating an embedding of the text content and metadata of documents, which is then used for various tasks such as similarity search and question answering. This uses the LanceDB vector store.
 
 ### AI-Enhanced Suggestions
 

src/paperless_ai/tests/test_ai_indexing.py

@@ -56,7 +56,7 @@ class FakeEmbedding(BaseEmbedding):
         return [0.1] * self.get_query_embedding_dim()
 
     def get_query_embedding_dim(self) -> int:
-        return 384  # Match your real FAISS config
+        return 384  # Match the test embedding dimension
 
 
 @pytest.mark.django_db
@@ -615,7 +615,7 @@ def test_query_similar_documents_empty_allow_list_fails_closed(
 class TestUpdateLlmIndexEmptyDocumentSet:
     """update_llm_index must persist an empty index when all documents are deleted.
 
-    Without this, the stale on-disk FAISS vectors are never cleared and
+    Without this, stale on-disk vector rows are never cleared and
     subsequent similarity searches return phantom hits for document IDs that
     no longer exist in the DB.
     """
@@ -731,7 +731,7 @@ class TestLlmIndexAddOrUpdateDocumentEmptyContent:
 
 @pytest.mark.django_db
 class TestLlmIndexLocking:
-    """The FAISS index mutation functions must acquire the index lock before touching the index.
+    """The LLM index mutation functions must acquire the index lock before touching the index.
 
     Without locking, two concurrent Celery workers can each load the same
     on-disk index, make independent modifications, and the last writer silently
@@ -834,7 +834,7 @@ class TestLlmIndexLocking:
 
         # exists=True so the code reaches the lock; iterate over an empty
         # queryset so VectorStoreIndex is called with no nodes (still exercises
-        # the lock path without needing heavy FAISS fixture data)
+        # the lock path without needing heavy vector-store fixture data)
         mock_qs = MagicMock()
         mock_qs.exists.return_value = True
         mock_qs.__iter__ = MagicMock(return_value=iter([]))