Ex-Googlers are building infrastructure to help companies understand their video data

As video content creation reaches unprecedented levels, companies are left with vast amounts of unutilized footage. From extensive broadcast archives to numerous surveillance cameras, much of this video remains dormant on servers, representing a wealth of unexplored information. This phenomenon, known as dark data, poses a challenge for businesses that collect it but rarely leverage its potential. To address this issue, Aza Kai and Hiraku Yanagita, former colleagues at Google Japan, have co-founded InfiniMind, a startup based in Tokyo. Their mission is to create infrastructure that transforms petabytes of unviewed video and audio into structured, accessible data for businesses. Kai noted that during their tenure at Google, they recognized the impending need for such a solution, which has now become increasingly apparent. With advancements in technology, particularly in vision-language models, video AI has evolved significantly from basic object recognition to more complex analyses. Kai explained that earlier methods struggled to provide insights into narratives or causality within video content. This limitation left clients with extensive archives unable to answer even fundamental questions about their footage. InfiniMind recently raised $5.8 million in seed funding, with contributions from UTEC, CX2, Headline Asia, Chiba Dojo, and an AI researcher at a16z Scout. The startup plans to move its headquarters to the United States while maintaining its Japanese office, which has served as a vital testing ground due to its robust technological infrastructure and supportive startup community. The company launched its first product, TV Pulse, in Japan in April 2025. This AI-driven platform analyzes television content in real time, enabling media and retail businesses to track product visibility, brand presence, customer sentiment, and public relations impact. After successful pilot programs with major broadcasters, InfiniMind has already secured paying clients. Looking ahead, InfiniMind is preparing to introduce its flagship product, DeepFrame, a long-form video intelligence platform capable of analyzing 200 hours of footage to identify specific scenes and events. Beta testing is set to begin in March, with a full launch anticipated in April 2026. The video analysis sector is crowded, with various companies offering general video understanding APIs. However, InfiniMind distinguishes itself by focusing on enterprise applications, such as monitoring and security, while integrating audio and speech analysis alongside visual data. Kai emphasized that their solution is user-friendly, requiring no coding, and that it can handle unlimited video lengths efficiently. The recent funding will allow InfiniMind to enhance its model, expand its engineering team, and reach new customers in both Japan and the U.S. Kai expressed enthusiasm for the potential of video intelligence, stating that it is a significant step towards achieving artificial general intelligence (AGI) by improving our understanding of reality and enabling better decision-making for humanity.