diff --git a/db.py b/db.py
index b0e8ecc..1d50c1e 100644
--- a/db.py
+++ b/db.py
@@ -35,8 +35,6 @@ class Database:
     which can then be used to obtain record.
     TODO For faster nearest neighbour lookup we should use something else,
          e.g. kd-trees
-    TODO The resulting db is huge (50 MB for 4 pdfs), contains a lot of duplicit uncompressed text.
-         We should at least de-duplicate the document path.
     """
     vectors: list[Vector]
     records: dict[VectorBytes, Record]