Memories.ai launches AI platform to analyze 10 million hours of video
Memories.ai is tackling the problem of AI struggling to understand long, multi-video contexts. Its platform can process up to 10 million hours of footage, offering searchable indexing, tagging, and aggregation across huge video libraries.
The startup targets security firms sifting through thousands of hours from multiple cameras and marketing teams studying video campaigns. 
Co-founders Dr. Shawn Shen and Enmin (Ben) Zhou both worked at Meta, bringing deep AI and ML expertise. Shen said traditional AI models hit limits beyond 1-2 hours of video.
“All top AI companies, such as Google, OpenAI and Meta, are focused on producing end-to-end models. Those capabilities are good, but these models often have limitations around understanding video context beyond one or two hours,” Shen told TechCrunch.
“But when humans use visual memory, we sift through a large context of data. We were inspired by this and wanted to build a solution to understand video across many hours better.”
The company just raised $8 million in seed funding. Susa Ventures led, with Samsung Next, Fusion Fund, Crane Ventures, Seedcamp, and Creator Ventures also investing. They had initially aimed for $4 million but saw heavy demand.
Susa partner Misha Gordon-Rowe praised Shen’s technical chops:
“Shen is a highly technical founder, and he is obsessed with pushing boundaries of video understanding and intelligence,” said Gordon-Rowe.
“Memories.ai can unlock a lot of first-party visual intelligence data with its solution. We felt that there was a gap in the market for long context visual intelligence, which attracted us to invest in the company.”
Samsung Next sees consumer potential in on-device computing that protects privacy by reducing cloud storage needs:
“One thing we liked about Memories.ai is that it could do a lot of on-device computing. That means you don’t necessarily need to store video data in the cloud. This can unlock better security applications for people who are apprehensive of putting security cameras in their house because of privacy concerns,” said Sam Campbell from Samsung Next.
The company’s tech first removes noise from video, compresses data, then uses indexing to make footage searchable via natural language queries. It also includes segmenting, tagging, and an aggregation layer to summarize findings.
Currently, they serve marketing teams to analyze video trends and create content, and security firms to detect suspicious behavior through pattern recognition.
Users upload video libraries now, but Shen said syncing and shared drives will come soon. The goal: users asking AI questions like “Tell me all about people I interviewed last week.”
Shen also sees the tech aiding humanoid robot training and self-driving cars by remembering routes better.
Memories.ai has 15 employees and plans to grow the team and sharpen its search tech with this funding.
Its competitors include mem0 and Letta, which focus on AI memory but with limited video features. Also, TwelveLabs and Google work on related video understanding. Shen says his approach is more horizontal, to work with multiple video models.
Image Credits: Memories.ai
Image Credits: Screenshot by TechCrunch
