Memories.ai Announced a Collaboration with Nvidia at its GTC Conference

Memories.ai, an Al company dedicated to reshaping machine perception and interaction, led by Dr. Shawn Shen and Enmin (Ben) Zhou, has announced a collaboration with semiconductor giant Nvidia to let Memories.ai use Nvidia’s Cosmos-Reason 2, a reasoning vision language model, and Nvidia Metropolis, an application for video search and summarization, to continue to develop its visual memory technology.

Shen and Ben conceived the company while developing the AI system behind Meta Ray-Ban smart glasses. The experience highlighted a key challenge: real-world usability would be limited if recorded video data could not be easily recalled.

After finding no existing solutions for visual memory in AI, Shen and Ben left Meta to build it themselves.

“AI performs strongly in the digital world, but the physical world remains largely untapped. Wearables and robotics require memory systems, particularly visual memory, to function effectively,” Shen noted, outlining the long-term vision.

AI memory capabilities have only recently begun to evolve. OpenAI introduced memory features in ChatGPT in 2024 and refined them in 2025, while Elon Musk’s xAI and Google’s Gemini launched similar tools.

However, these advancements have largely focused on text-based memory, structured and easier to index, making them less suitable for physical AI systems that rely on visual understanding of the world.

Memories.ai launched in 2024 and has raised $16 million to date, including an $8 million seed round in July 2025 and an $8 million extension led by Susa Ventures, with participation from Seedcamp, Fusion Fund, and Crane Venture Partners.

Building a visual memory layer required two core elements: infrastructure to embed and index video into retrievable formats, and large-scale data collection to train the models.

The company introduced its large visual memory model (LVMM) in July 2025, positioned as a smaller counterpart to Gemini Embedding 2. To support training, a proprietary hardware device called LUCI was developed to capture video data, addressing limitations in off-the-shelf recorders.

A second-generation LVMM has since been released, alongside a partnership with Qualcomm to run on its processors. While collaborations with major wearable companies are underway, the primary focus remains on advancing the model and infrastructure, with larger commercialization opportunities expected as the wearables and robotics markets mature.