The RelGT-AC model enhances autocomplete tasks in relational databases by utilizing a novel column masking strategy and a TF-IDF text encoder, significantly improving predictive performance. This advancement in relational deep learning could streamline data entry processes across various sectors, including enterprise and healthcare.