Memories.ai and Nvidia Transform AI Visual Memory

⚡

Key Takeaways

1Memories.ai is collaborating with Nvidia to develop a visual memory for AI, integrating Cosmos-Reason 2 and Metropolis.

2Founded by Shawn Shen and Ben Zhou, the company has raised $16 million for its visual memory technology.

3The LVMM model from Memories.ai, launched in 2025, works with Qualcomm processors and targets wearables and robotics.

💡Why it matters — This advancement could transform how wearable devices and robots interact with the physical world, enhancing their ability to remember visually.

A New Era for Visual Memory in AI

Shawn Shen, founder of Memories.ai, believes that for artificial intelligence (AI) to succeed in the physical world, it must be able to remember what it sees. With this in mind, Memories.ai, his company, has partnered with Nvidia to develop an infrastructure that allows wearable devices and robotics to store and recall visual memories. This collaboration leverages Nvidia's advanced AI tools.

A Strategic Partnership with Nvidia

At Nvidia's GTC conference, Memories.ai announced its partnership with the semiconductor giant. This partnership enables Memories.ai to utilize Nvidia's reasoning vision language model, Cosmos-Reason 2, as well as the Metropolis application, dedicated to research and video synthesis. These tools are essential for the ongoing development of Memories.ai's visual memory technology.

Origins in Meta's Smart Glasses

Shawn Shen, along with his co-founder Ben Zhou, came up with the idea for Memories.ai while working on the AI system for Meta's Ray-Ban glasses. This experience led them to consider the everyday use of this technology, particularly whether users could remember recorded videos. Finding no existing solution, they decided to leave Meta to develop their own visual memory solution.

The Importance of Visual Memories for AI

Shen emphasizes that while AI has already proven itself in the digital realm, it still needs to adapt to the physical world. For this, wearable devices and robotics require visual memory. Shen and his team firmly believe in a future where AI can remember visually, which is crucial for its integration into the physical world.

Recent Advances in AI Memory

The ability of AI systems to remember is still relatively new. OpenAI introduced memory features in ChatGPT as early as 2024, and other companies like Elon Musk's xAI and Google Gemini have also launched similar tools. However, these innovations primarily focus on textual memory, which is more structured but less suited for physical applications requiring visual memory.

The Financial Journey of Memories.ai

Founded in 2024, Memories.ai has raised $16 million to date. In July 2025, it completed a funding round of $8 million, followed by an additional $8 million extension. This funding was led by Susa Ventures and included participation from Seedcamp, Fusion Fund, and Crane Venture Partners.

Technical Challenges and Innovations

To successfully create effective visual memory, Memories.ai must build an infrastructure capable of integrating and indexing videos in a storable and retrievable data format. Additionally, capturing the necessary data to train the model to accomplish this task is essential.

The Large Visual Memory Model (LVMM)

In July 2025, Memories.ai launched its Large Visual Memory Model (LVMM), comparable to a more compact version of Gemini Embedding 2. This multimodal indexing and retrieval model was introduced earlier this year.

Data Collection and the LUCI Device

To gather the necessary data, Memories.ai developed LUCI, a hardware device worn by its "data collectors" that records videos to train the model. While the company does not plan to become a hardware manufacturer, it designed LUCI to address the limitations of current video recorders, which are often too power-hungry and high-definition.

Collaboration with Qualcomm and Future Prospects

Memories.ai recently released the second generation of its LVMM and signed a partnership with Qualcomm to ensure its model runs on Qualcomm processors by the end of the year. The company is also collaborating with several major wearable device companies, although Shen has not disclosed their names. Despite current demand, Shen sees even greater potential in wearables and robotics in the future.

A Long-Term Vision for Wearables and Robotics

Shen states that while immediate commercialization is not their priority, they are focused on developing the model and infrastructure. He is confident that the market for wearables and robotics will eventually emerge, even if it takes time.