Memories AI Pioneers Visual Memory Technology with $8M Funding

San Francisco, July 25, 2025 - Memories AI, an emerging artificial intelligence research lab founded by former Meta and Cambridge University researchers, today announced the launch of the world's first Large Visual Memory Model (LVMM) alongside an $8 million seed funding round led by Susa Ventures.

Breakthrough in AI Memory Capabilities

The company's proprietary LVMM technology represents a significant leap forward in machine perception. Unlike conventional AI systems limited to processing short video clips (15-60 minutes), Memories AI's architecture can analyze and retain visual information across millions of hours of footage - mirroring human memory functions.

"Our technology answers fundamental questions current AI cannot," explained the company's CEO. "Systems can now recall whether they've seen something before or identify changes over time - capabilities critical for true contextual understanding."

The three-layer architecture works by:

1. Processing: Denoising and compressing video while extracting key elements
2. Indexing: Creating searchable references accessible via natural language queries
3. Consolidation: Structuring visual data for pattern recognition and temporal comparisons

Cross-Industry Applications

The LVMM platform demonstrates remarkable versatility across multiple sectors:

• Security: Enables anomaly detection through continuous surveillance analysis
• Marketing: Tracks brand mentions and consumer trends across social video platforms
• Autonomous Systems: Provides robots and vehicles with persistent environmental memory

The technology is accessible through API integration or a chatbot interface, allowing users to query video content using everyday language.

Investor Confidence Fuels Expansion

The oversubscribed seed round - doubling its initial $4 million target - included participation from Samsung Next, Crane Venture Partners, and other prominent investors. Funding will accelerate:

• Engineering team growth
• Privacy framework development
• Enterprise client acquisition

"This isn't just another computer vision solution," noted a Susa Ventures partner. "Memories AI is building foundational infrastructure that could enable breakthroughs from robotics to AGI."

Founding Team Brings Meta Expertise

The startup benefits from its founders' deep experience in Meta's Reality Labs division, where they pioneered visual AI systems. Their specialized knowledge in multimodal learning and device-side processing directly informed LVMM's innovative architecture.

Key Points:

• World-first visual memory model processes millions of video hours
• 100x capacity improvement over existing technologies
• $8M funding to expand commercial applications
• Horizontal architecture compatible with multiple video models
• Founding team combines Meta research experience with Cambridge academic rigor