Modeling tools have become irreplaceable tools that are widely used to understand seawater intrusion processes. With the increasing usage of Artificial intelligence in all aspects of scientific applications and the incredible results, seawater intrusion could be one of the topics that AI, and machine learning could significantly help. Machine learning can provide reliable predictions by learning from available datasets. Despite the numerous advantages of the machine learning models, there is one major drawback, they rely on a vast amount of data, which is often costly and unreachable in several scientific domains, in hydrogeology. In response to these challenges, physics-informed neural networks (PINNs) have been implemented as an innovative approach that integrates physical laws directly into the training process of neural networks. Unlike conventional machine learning models that depend solely on data, PINNs incorporate governing mathematical equations, such as differential equations into the network’s loss function. To the best of our knowledge, PINNs approach has never been applied to seawater intrusion. This study addresses this gap and aims at evaluating the performance of PINNs for seawater intrusion in coastal aquifers. The Henry problem, a widely used hypothetical model, describing seawater intrusion in coastal aquifers, is employed in this context. The study's results showed that standard physics-informed neural networks successfully captured crucial aspects of saltwater intrusion, highlighting the effectiveness of this method in hydrogeological modeling. While this method is effective for modeling seawater intrusion using PINNs, it faces challenges when dealing with complex problems, especially in transitioning from a wider to a narrow mixing zone. A key finding of this research is that transfer learning, applying knowledge from simpler cases to more complex ones, can help improve results.