Samsung is investing in Memories.ai, an early-stage startup that builds powerful video analytics tools.
The company recently raised $8 million in seed funding led by Susa Ventures, and has earned participation from Samsung Next, Fusion Fund, Crane Ventures, Seedcamp and Creator Ventures.
Startup Sales Points provide AI tools that can process and understand up to 10 million hours of videos.
This solves the main problem. To make long format videos searchable and convenient.
Today's Video AI
Most video AI tools today can handle short clips. You can detect objects and summarise scenes. But they struggle with footage that lasts for hours or multiple days.
This is a major limitation for sectors like security and marketing. The security team is tasked with checking the time of surveillance footage from various cameras.
Marketers often analyze numerous video campaigns to identify trends, but without the right tools, both tasks require long manual reviews.
memories.ai remedies by providing searchable indexing, tagging, segmentation, and video data aggregation.
Users can ask complex questions on large video datasets and receive relevant context-enabled results.
Experienced founder
The company has strong technical experience in its leadership. Co-founder Dr. Shawn Shen worked as a research scientist at Meta's Reality Lab while pursuing his PhD.
His co-founder, Enmin (Ben) Zhou, was also in Meta. I worked there as a machine learning engineer.
According to Shen, most AI companies today focus on building end-to-end models. These models are useful, but their ability to understand long format video contexts is limited.
“In contrast, humans process visual information over time,” Shen explained. “We wanted to create an AI that could do the same thing.”
structure
Memories.ai uses a technology stack and processes the video in several steps. First removes noise from the video and saves important moments.
Next, we index the sieve moments and provide a summary. The result is a system that allows users to understand long videos and search for them using plain language queries.
For example, users can ask, “Show me last week's interview,” and the system will deliver accurate results.
Use Cases
At the moment, Memories.ai covers two industries: marketing and security. Marketers use the platform to analyse social media videos, review past campaigns, and identify content trends.
It also includes basic tools for creating videos. Meanwhile, security companies use it to check surveillance footage. AI can detect patterns, recognize suspicious behavior, and help teams act faster.
Read again: Why AI Security Starts with Access Control
Samsung's interest
Samsung Next, the company's investment arm, sees significant potential. It emphasizes memory.
This means that users may not need to upload footage to the cloud. According to Sam Campbell, a partner at Samsung Next, this could solve consumer privacy concerns.
Many people are reluctant to use home security cameras because they are worried about storing video data online. Edge computing offers a more private alternative.
Future plans
Today, businesses need to upload video archives for analysis. However, the startup plans to introduce shared drives and real-time syncs. This makes it easier for users to manage their content.
The team will also be building AI assistants. This assistant can help users remember information and automate tasks using personal videos, photos, or smart glasses.
In the long term, technology can support applications such as robot training and enhanced navigation in self-driving cars.
competition
Memories.ai faces competition with startups like MEM0 and Letta. These companies are also working on the video memory layer of AI, but they have limited products.
Large companies such as Twelvelabs and Google are also active in this field. However, the Memories.ai solution looks more promising.
The system can support many types of AI models and serve multiple industries.
