The article introduces CAR-Toner, an AI-driven tool for predicting and optimizing the tonic signaling of chimeric antigen receptors (CARs). Tonic signaling is crucial for CAR-T cell fitness, affecting both persistence and exhaustion. The authors previously developed a bioinformatic method to calculate positively charged patches (PCPs) on the surface of CARs, which are indicative of tonic signaling. However, this method had limitations, including reliance on external servers, long calculation times, lack of batch processing, and no optimization strategies. To address these issues, CAR-Toner integrates protein databases, structural biology, and advanced deep learning models. It uses a comprehensive protein structure database and an in-house algorithm to calculate PCP scores. The ESM2 model, a transformer-based language model, is used for fine-tuning tasks. CAR-Toner can efficiently calculate PCP scores for individual proteins, process multiple sequences in batches, and optimize PCP scores. CAR-Toner demonstrated high accuracy in predicting PCP scores and effectively tuned PCP scores of various CAR variants. It was also used to optimize a camelid single-domain nanobody (VHH) against a tumor-associated antigen, showing improved CAR-T cell function. The tool was further applied to compare PCP features of different CAR antigen-binding domains, revealing that VHH-based CARs have the highest proportion of optimal PCP scores. Overall, CAR-Toner is a powerful tool for enhancing CAR-T design by optimizing tonic signaling, paving the way for advancements in CAR-T therapy.The article introduces CAR-Toner, an AI-driven tool for predicting and optimizing the tonic signaling of chimeric antigen receptors (CARs). Tonic signaling is crucial for CAR-T cell fitness, affecting both persistence and exhaustion. The authors previously developed a bioinformatic method to calculate positively charged patches (PCPs) on the surface of CARs, which are indicative of tonic signaling. However, this method had limitations, including reliance on external servers, long calculation times, lack of batch processing, and no optimization strategies. To address these issues, CAR-Toner integrates protein databases, structural biology, and advanced deep learning models. It uses a comprehensive protein structure database and an in-house algorithm to calculate PCP scores. The ESM2 model, a transformer-based language model, is used for fine-tuning tasks. CAR-Toner can efficiently calculate PCP scores for individual proteins, process multiple sequences in batches, and optimize PCP scores. CAR-Toner demonstrated high accuracy in predicting PCP scores and effectively tuned PCP scores of various CAR variants. It was also used to optimize a camelid single-domain nanobody (VHH) against a tumor-associated antigen, showing improved CAR-T cell function. The tool was further applied to compare PCP features of different CAR antigen-binding domains, revealing that VHH-based CARs have the highest proportion of optimal PCP scores. Overall, CAR-Toner is a powerful tool for enhancing CAR-T design by optimizing tonic signaling, paving the way for advancements in CAR-T therapy.