Traditional pharmacokinetic tools, such as NONMEM (Nonlinear Mixed Effects Modeling), have long been the standard for modeling drug concentration over time. However, these tools face significant limitations, including high costs, a steep learning curve, and reliance on outdated computational techniques, making them inaccessible to smaller medical institutions. To address these challenges, we propose Pharmaformer, a Transformer-based pharmacokinetic prediction system. Leveraging the advanced capabilities of Transformer models in capturing complex temporal patterns, Pharmaformer enables precise and interpretable predictions of drug concentrations. The system integrates data preprocessing, model inference, and results output into a user-friendly and cost-effective framework. Experimental results demonstrate that Pharmaformer outperforms the traditional FOCE algorithm, which follows the same principles as the one in NONMEM, in key metrics, offering a modern and accessible alternative for pharmacokinetic modeling. This system lowers barriers to entry while enhancing prediction accuracy, offering clinicians and researchers an innovative tool for improved pharmacokinetic analysis.