The Article Tells the Story of:
- Samsung Invests in AI Video Startup: Memories.ai raised $8 million to scale long-context video analysis.
- Startup Promises On-Device Privacy and Searchable Indexing: It uses custom models to compress, tag, and summarize video data.
- Target Markets Are Security and Marketing: Clients use the AI to study hours of footage or social trends.
- Google and OpenAI Face Competition: Memories.ai competes by focusing on long-form video context.
Samsung Backs Memories.ai to Handle Long-Form Video Analysis
At Squaredtech.co, we track how AI is expanding into real-world applications. One of the latest moves comes from Samsung, which just backed a video AI startup called Memories.ai. The company focuses on helping businesses process and analyze massive video libraries with minimal storage and maximum context.
Memories.ai raised $8 million in seed funding in a round led by Susa Ventures. Other participants included Samsung Next, Fusion Fund, Crane Ventures, Seedcamp, and Creator Ventures. The company originally planned to raise $4 million but exceeded its goal due to investor interest.
Read More About Our Article of MiniMax: The Game-Changing AI Video Generator Making Waves in China. Published on September 4th, 2024 SquaredTech
Samsung sees value in Memories.ai’s ability to analyze video directly on devices, which could reduce reliance on cloud storage. This feature could appeal to consumers worried about privacy, especially when it comes to home security systems.
Memories.ai’s platform is built to understand up to 10 million hours of video. That’s a level of long-term, multi-source context that typical AI video models—like those from Google or OpenAI—don’t support. Many standard AI tools summarize one video or analyze a few minutes of footage. But Memories.ai is aiming much higher, offering companies an AI that can index, tag, and summarize entire video libraries.
How Memories.ai Breaks Down Massive Video Libraries
Memories.ai uses a custom-built AI stack that includes compression, indexing, segmentation, and aggregation layers:
- Compression Layer: Strips away noise and keeps only the relevant parts of each video.
- Indexing Layer: Makes videos searchable using natural-language queries.
- Segmentation: Breaks videos into useful chunks and adds metadata tags.
- Aggregation: Summarizes results to create shareable reports.
Co-founder Dr. Shawn Shen, a former Meta Reality Labs researcher, explains the motivation behind the platform:
“All major AI companies focus on short-term end-to-end models. But real human memory uses longer visual context. We built a system to reflect that.”
Shen’s co-founder Enmin (Ben) Zhou also worked at Meta as a machine learning engineer. Together, they developed Memories.ai to handle long video contexts—something that traditional AI models struggle to manage.
Right now, companies upload their video libraries directly into the platform. But Shen says the team is building features to sync content automatically through a shared drive, allowing businesses to query their footage with commands like:
“Show me everyone I interviewed last week.”
At Squaredtech.co, we see this approach as a major step toward making AI-driven video analysis more interactive and scalable for real use cases.
Security and Marketing Companies Are the First Clients
Memories.ai currently serves two main markets:
- Security companies use the AI to detect patterns of suspicious activity across hours of surveillance footage.
- Marketing teams use it to study branded video content, understand audience engagement, and even plan future campaigns.
For marketers, the platform helps track trends across platforms like YouTube or Instagram. It also offers basic tools to help generate videos once insights are collected.
The startup also imagines more advanced use cases in the future. Shen says the long-context AI could help:
- Train humanoid robots by feeding them video of real tasks.
- Support self-driving cars by letting them remember previous routes.
- Act as a personal assistant, using smart glasses to process and recall life events.
Memories.ai plans to grow its current 15-person team and improve its video search capabilities using the new funding.
Samsung and Other Investors See Broad Potential
At Squaredtech.co, we asked why Samsung chose to invest. The answer is simple: on-device AI with privacy benefits.
Sam Campbell, partner at Samsung Next, explained that Memories.ai can process data without uploading everything to the cloud. That means users can use security systems or video tools without worrying about cloud-based surveillance or privacy leaks.
This fits Samsung’s vision of smart, privacy-conscious devices that still provide rich AI-driven experiences. The move suggests that Memories.ai may soon see use cases in mobile phones, smart home devices, or wearables.
Investor Misha Gordon-Rowe from Susa Ventures also pointed to a market gap in long-form video analysis as a reason for their investment.
“Memories.ai fills a clear gap in the market. No one else is offering true visual intelligence across long timelines.”
Competition Is Rising in the Video AI Space
Memories.ai isn’t alone in this space. Other startups like mem0 and Letta also aim to give AI systems memory layers. However, most of them don’t yet support large-scale video libraries. They focus more on short clips or image-based memories.
Meanwhile, tech giants like Google and OpenAI are improving their video understanding tools. But these typically stop at processing one or two hours of footage. Shen says Memories.ai can work across different models and offer a horizontal layer that supports multiple AI workflows.
By building an independent pipeline that compresses, tags, and summarizes long videos, Memories.ai separates itself from other solutions that rely entirely on large foundational models.
At Squaredtech, we think this horizontal flexibility—paired with direct indexing and on-device support—gives Memories.ai a technical edge.
Final Thoughts from Squaredtech
Memories.ai isn’t just another AI startup—it’s targeting a major gap in video intelligence. Companies are swimming in content, and current AI tools can’t keep up. With custom layers that compress, tag, and summarize video, Memories.ai is trying to make video analysis as searchable as text.
Samsung’s investment signals confidence in the product’s future. Whether it’s for home security, marketing research, or AI assistants, this startup is tackling a problem that even Google hasn’t solved yet.
At Squaredtech.co, we’ll be watching closely as Memories.ai expands its platform, scales up its team, and possibly brings its tech to Samsung’s own smart devices.
This is more than just another seed round. This could shape how future AI tools remember, search, and understand visual data.
Stay Updated: Artificial Intelligence
