Echocardiography is the first-line imaging technique in the diagnosis and treatment of cardiovascular diseases. However, it has limitations such as high operator dependence and long inspection time. Artificial intelligence ( AI) technology is rapidly transforming the clinical application of echocardiography. The application of AI-based echocardiography involves multiple levels and stages of the echocardiography workflow, as well as different application scenarios. These include the image acquisition, quality assessment, automatic measurement of image analysis, feature extraction and systematic report generation. Existing studies demonstrate that AI can significantly improve the image acquisition efficiency, interpretability and diagnostic accuracy. Nevertheless, the application of AI in echocardiography still faces numerous challenges such as poor quality of raw image datasets, " black-box" characteristics and potential security vulnerabilities. These issues can be progressively solved through the development of hybrid or multimodal AI models, the establishment of large-scale, diverse, high-quality image-text database and the employment of federated learning techniques. Thereby, it can achieve the full automation of echocardiographic analysis, enhance the diagnostic accuracy and examination efficiency for cardiovascular diseases and optimize the clinical workflows.