AbNatiV is a deep learning tool for assessing the nativeness of antibodies and nanobodies, determining their likelihood of belonging to the distribution of immune-system-derived human or camelid antibodies. It accurately predicts the nativeness of Fv sequences from any source, including synthetic libraries and computational design. AbNatiV provides an interpretable score predicting immunogenicity and a residue-level profile guiding antibody and nanobody engineering. It also includes an automated humanization pipeline, which was applied to two nanobodies. Laboratory experiments showed that AbNatiV-humanized nanobodies retained binding and stability at par or better than their wild type, unlike those humanized using conventional methods. AbNatiV is available as downloadable software and a webserver. Antibodies are biomolecules that can bind to molecular targets selectively and tightly, making them key in biological research and medicine. Nanobodies (Nb), single-domain antibodies naturally expressed in camelids, have gained popularity due to their unique structural characteristics. Since the approval of the first nanobody drug, Caplacizumab, in 2019, their potential as therapeutics has increased. Traditional methods for discovering antibodies or nanobodies include in vivo approaches like animal immunization and in vitro techniques like library construction and screening. Computational design has emerged as a third generation approach. In vitro methods like phage display have shown promise to replace animal immunization. However, immune-system-derived antibodies tend to have better in vivo properties, including long half-life, low immunogenicity, and low toxicity. Despite this, computational design remains in its infancy, with important advances in antibody design targeting specific epitopes and predicting biophysical properties. However, computational prediction of in vivo properties remains challenging. AbNatiV addresses this by enabling computational engineering of antibody and nanobody sequences indistinguishable from those obtained from immune systems. AbNatiV uses a VQ-VAE architecture trained on immune-system-derived sequences, providing an interpretable score and residue-level profile. It was tested on human and camelid sequences, showing high accuracy in distinguishing native from non-native sequences. AbNatiV outperforms alternative methods in classification tasks and has higher reconstruction accuracy. It was applied to two nanobodies, showing that AbNatiV-humanized nanobodies retained binding and stability at par or better than their wild type. AbNatiV is available as downloadable software and a webserver.AbNatiV is a deep learning tool for assessing the nativeness of antibodies and nanobodies, determining their likelihood of belonging to the distribution of immune-system-derived human or camelid antibodies. It accurately predicts the nativeness of Fv sequences from any source, including synthetic libraries and computational design. AbNatiV provides an interpretable score predicting immunogenicity and a residue-level profile guiding antibody and nanobody engineering. It also includes an automated humanization pipeline, which was applied to two nanobodies. Laboratory experiments showed that AbNatiV-humanized nanobodies retained binding and stability at par or better than their wild type, unlike those humanized using conventional methods. AbNatiV is available as downloadable software and a webserver. Antibodies are biomolecules that can bind to molecular targets selectively and tightly, making them key in biological research and medicine. Nanobodies (Nb), single-domain antibodies naturally expressed in camelids, have gained popularity due to their unique structural characteristics. Since the approval of the first nanobody drug, Caplacizumab, in 2019, their potential as therapeutics has increased. Traditional methods for discovering antibodies or nanobodies include in vivo approaches like animal immunization and in vitro techniques like library construction and screening. Computational design has emerged as a third generation approach. In vitro methods like phage display have shown promise to replace animal immunization. However, immune-system-derived antibodies tend to have better in vivo properties, including long half-life, low immunogenicity, and low toxicity. Despite this, computational design remains in its infancy, with important advances in antibody design targeting specific epitopes and predicting biophysical properties. However, computational prediction of in vivo properties remains challenging. AbNatiV addresses this by enabling computational engineering of antibody and nanobody sequences indistinguishable from those obtained from immune systems. AbNatiV uses a VQ-VAE architecture trained on immune-system-derived sequences, providing an interpretable score and residue-level profile. It was tested on human and camelid sequences, showing high accuracy in distinguishing native from non-native sequences. AbNatiV outperforms alternative methods in classification tasks and has higher reconstruction accuracy. It was applied to two nanobodies, showing that AbNatiV-humanized nanobodies retained binding and stability at par or better than their wild type. AbNatiV is available as downloadable software and a webserver.