You start by putting in a question or concept into Sparkify’s interface. The Gemini model, a large language model, interprets the input, extracting key information and generating a coherent narrative. Simultaneously, the Veo model, designed for video generation, creates visual elements such as animations, charts, and motion grafics.
The result is a concise video, typically under two minutes, that combines text explanations, visuals, and AI-generated voiceovers to present the information in an easily accessible format, mostly on the level of edutainment or childhood education.  Sparkify integrates multiple modes of information to explain. For instance, for a scientific concept like electromagnetic induction, Sparkify can produce an animation illustrating how a changing magnetic field induces current in a circuit, accompanied by verbal explanation.
This multimodal approach improves comprehension and retention, making it a valuable resource for educators and students.
Currently, Sparkify is in an experimental phase. If you want to try it, you need to join a waiting list. Once accepted you can input queries or concepts, select preferred video styles and lengths, and receive animated videos generated by the AI models and all that without any experience in animation or video production.