Samsung backs a video AI startup that can analyze thousands of hours of footage

In a significant move for the realm of video analysis, Samsung has thrown its support behind Memories.ai, a startup that promises to change how businesses process vast amounts of video content. While existing AI tools can summarize single videos effectively, they often struggle with analyzing data across multiple hours—an issue that poses challenges for security firms needing to sift through extensive footage from various sources, as well as marketers examining diverse video campaigns. Memories.ai is setting out to address this challenge with its advanced AI platform, capable of processing an astonishing 10 million hours of video. The startup aims to provide a contextual framework that includes searchable indexing, tagging, segmentation, and aggregation, allowing companies to extract meaningful insights from their video libraries. Co-founder Dr. Shawn Shen, who previously worked as a research scientist at Meta’s Reality Labs, shared that the inspiration for the platform came from the way humans utilize visual memory to navigate large datasets. In a recent seed funding round led by Susa Ventures, Memories.ai secured $8 million, significantly surpassing its initial goal of $4 million due to overwhelming investor interest. Misha Gordon-Rowe, a partner at Susa Ventures, praised Shen's technical expertise and ambition to push the boundaries of video understanding, highlighting the market gap for long-context visual intelligence that Memories.ai aims to fill. Samsung Next, the investment arm of Samsung, recognizes the consumer applications of Memories.ai's technology. Partner Sam Campbell pointed out the platform's capability for on-device computing, which alleviates concerns about privacy and data storage in the cloud—a critical factor for consumers wary of security cameras at home. The startup employs a unique tech stack that enhances video analysis. First, it filters out noise from the footage before compressing the data to retain only essential information. This is followed by an indexing process that allows for natural-language queries, segmentation, and tagging, culminating in an aggregation stage that creates reports from the indexed data. Currently, Memories.ai caters to two primary sectors: marketing and security. Marketing organizations can leverage the platform to discern social media trends and strategize video content, while security companies can utilize its tools to analyze surveillance footage for detecting potential threats. Looking ahead, Shen envisions a future where clients can seamlessly sync video content and engage in intuitive queries about their footage. He also imagines the technology assisting in training humanoid robots and enhancing the memory capabilities of self-driving cars. With a growing team of 15, Memories.ai plans to use the recent funding to expand its workforce and refine its search capabilities. As it navigates the competitive landscape filled with similar startups and tech giants like Google, Shen is confident that Memories.ai's comprehensive approach will enable it to collaborate effectively with various video models, setting it apart in the rapidly evolving field of video intelligence.