diff --git a/damus-c/cursor.h b/damus-c/cursor.h
index 96427535..bb3cc30f 100644
--- a/damus-c/cursor.h
+++ b/damus-c/cursor.h
@@ -369,6 +369,20 @@ static inline int push_sized_str(struct cursor *cursor, const char *str, int len
 	return cursor_push(cursor, (u8*)str, len);
 }
 
+static inline int cursor_push_lowercase(struct cursor *cur, const char *str, int len)
+{
+	int i;
+
+	if (unlikely(cur->p + len >= cur->end))
+		return 0;
+
+	for (i = 0; i < len; i++)
+		cur->p[i] = tolower(str[i]);
+
+	cur->p += len;
+	return 1;
+}
+
 static inline int cursor_push_str(struct cursor *cursor, const char *str)
 {
 	return cursor_push(cursor, (u8*)str, (int)strlen(str));
diff --git a/nostrdb/nostrdb.c b/nostrdb/nostrdb.c
index 37ace98d..39167645 100644
--- a/nostrdb/nostrdb.c
+++ b/nostrdb/nostrdb.c
@@ -164,8 +164,8 @@ static int ndb_make_text_search_key(unsigned char *buf, int bufsize,
 	if (!push_varint(&cur, word_len))
 		return 0;
 
-	// non-null terminated string
-	if (!cursor_push(&cur, (unsigned char*)str, word_len))
+	// non-null terminated, lowercase string
+	if (!cursor_push_lowercase(&cur, str, word_len))
 		return 0;
 
 	// the index of the word in the content so that we can do more accurate