In AI age, the virtual and intelligent avater software applications represented by Talking Tom increasingly attract more and more attention. From an educational perspective, they can provide company for children, auxiliary learning and other functions. However, constrained by the complexity of related technologies, hardware and networks, it is very difficult for average education researchers to develop AI avatar applications which can offer rich functions and friendly user experience. Aiming at the problem, we propose a solution to developing a framework for AI avatar applications rapidly. It integrates the multiple cloud services from thirdparty providers, and implements virtual 3D avatar with human face synthesis, speech conversion, speech evaluation and social sharing. System testing and comparison with other related works show that the framework is suitable for common Android mobile devices and has the good response speed and friendly user experience, thus it becomes an auxiliary education teaching tool.