Businesses are producing more video than ever before. From years of broadcast archives to thousands of store cameras and countless production footage, much of it sits unused on our servers, unwatched and unanalyzed. This is dark data. It is a large, untapped resource that companies automatically collect but rarely use in any meaningful way.
To tackle this problem, Aza Kai (CEO) and Hiraku Yanagita (COO), two former Googlers who had worked together at Google Japan for nearly a decade, decided to build their own solution. The two co-founded InfiniMind, a Tokyo-based startup that develops infrastructure to transform petabytes of unwatched video and audio into structured, queryable business data.
“My co-founder has been leading brand and data solutions at Google Japan for 10 years, and I saw this tipping point coming even when I was still at Google,” Kai said. By 2024, the technology had matured and the market demand was clear enough that the co-founders felt they needed to start their own company, he added.
Kai, who previously worked on cloud, machine learning, advertising systems, and video recommendation models at Google Japan, and then led the data science team, explained that current solutions require trade-offs. Previous approaches could label objects within individual frames, but they were unable to track narratives, understand cause-and-effect relationships, or answer complex questions about video content. For clients with decades of broadcast archives and petabytes of footage, even basic questions about content were often unanswerable.
What has really changed is the advancement of visual language models from 2021 to 2023. Kai noted that this is when video AI began to move beyond simple object tagging. It helped that GPU costs had come down over the past decade, increasing annual performance by roughly 15% to 20%, but the bigger story was features. Until recently, he told TechCrunch, models couldn’t do that job.
InfiniMind recently secured $5.8 million in seed funding led by UTEC with participation from AI researchers from CX2, Headline Asia, Chiba Dojo, and a16z Scout. While the company will relocate its headquarters to the United States, it will continue to operate offices in Japan. Japan provided the perfect testbed with powerful hardware, talented engineers, and a supportive startup ecosystem, allowing the team to fine-tune the technology for demanding customers before expanding globally.
The first product, TV Pulse, was launched in Japan in April 2025. The AI-powered platform analyzes TV content in real-time, helping media and retail companies “track product exposure, brand presence, customer sentiment, and PR effectiveness,” the startup says. After pilot programs with major broadcasters and agencies, the company has already gained paying customers, including wholesalers and media companies.
tech crunch event
boston, massachusetts
|
June 23, 2026
InfiniMind is now ready for international markets. The company’s flagship product, DeepFrame, is a long-form video intelligence platform that can process 200 hours of footage to pinpoint specific scenes, speakers, and events, and will be released in beta in March, followed by a full release in April 2026, Cai said.
The video analytics space is highly fragmented. While companies such as TwelveLabs offer general-purpose video understanding APIs to a wide range of users, including consumers, prosumers and enterprises, Kai said InfiniMind is specifically focused on enterprise use cases such as monitoring, safety, security and analyzing video content for deeper insights.
“Our solution requires no code. Clients bring in their data and our systems process it to provide actionable insights,” said Cai. “We integrate not only visuals but also audio, sound, and speech understanding. Our system can handle unlimited video lengths, and cost efficiency is a major differentiator. Most existing solutions prioritize accuracy or specific use cases, but do not solve the cost challenge.”
The seed funding will help the team continue developing DeepFrame models, expand its engineering infrastructure, hire more engineers, and reach additional customers in Japan and the United States.
“This is an exciting space and one of the paths to AGI,” Kai said. “Understanding general video intelligence means understanding reality. Industrial applications are important, but our ultimate goal is to push the boundaries of technology to better understand reality and help humans make better decisions.”
