Many AI tools today can analyze a video and summarize its content, but they struggle when asked to process multiple videos or footage spanning many hours. This limitation poses a challenge for security companies that need to review thousands of hours of footage from various cameras, as well as marketing firms aiming to study different video campaigns and product shoots.
Memories.ai aims to address this problem with its AI platform, capable of processing up to 10 million hours of video. The startup provides a contextual layer for companies with large video libraries, offering searchable indexing, tagging, segmentation, and aggregation. Its co-founder, Dr. Shawn Shen, was a research scientist at Meta’s Reality Labs during his PhD, while his counterpart, Enmin (Ben) Zhou, worked as a machine learning engineer at Meta.
Shen explained that while top AI companies like Google, OpenAI, and Meta focus on end-to-end models, these often lack the ability to understand video context beyond one or two hours. Inspired by human visual memory, which processes vast amounts of data, Memories.ai was built to better comprehend video across extended durations.
The company recently raised $8 million in a seed funding round led by Susa Ventures, with participation from Samsung Next, Fusion Fund, Crane Ventures, Seedcamp, and Creator Ventures. Initially targeting $4 million, the round was oversubscribed due to strong investor interest.
Misha Gordon-Rowe, a partner at Susa Ventures, highlighted Shen’s technical expertise and obsession with advancing video understanding. He noted that Memories.ai fills a market gap by unlocking first-party visual intelligence data with long-context capabilities.
Samsung Next saw potential in the startup’s on-device computing approach, which avoids storing video data in the cloud. This feature could appeal to privacy-conscious consumers hesitant about security cameras in their homes.
Memories.ai uses its own tech stack and models for analysis. The process begins with noise removal and compression to retain only essential data. An indexing layer then makes videos searchable using natural-language queries, while an aggregation layer summarizes indexed data for reporting.
Currently, the startup serves marketing and security firms. Marketing teams can track brand trends on social media and identify video content strategies, while security companies use the platform to analyze footage for potential threats by detecting patterns in behavior.
Clients currently upload their video libraries for analysis, but future updates will allow syncing via shared drives. Shen envisions an AI assistant that gains context from users’ photos or smart glasses, with applications in training humanoid robots or aiding self-driving cars in route memorization.
With 15 employees, Memories.ai plans to expand its team and enhance its search capabilities. It competes with startups like mem0 and Letta, which focus on AI memory layers but offer limited video support, as well as companies like TwelveLabs and Google, which specialize in video understanding. Shen believes his company’s horizontal approach gives it an edge, enabling compatibility with various video models.