Memories.ai partners with Nvidia to build visual memory for AI

The collaboration aims to address the gap between digital and physical AI capabilities by building a visual memory layer. Memories.ai announced this partnership with Nvidia on Monday at GTC (Nvidia’s developer conference). Under the agreement, the startup will use Nvidia’s AI tools to develop its visual memory technology. Specifically, Memories.ai is utilizing Nvidia’s Cosmos-Reason 2, a reasoning vision language model, and Nvidia Metropolis, a reference architecture for video search and summarization, to build a visual memory layer for wearables and robotics. The company was launched in 2024 by co-founders Shawn Shen, who serves as CEO, and Ben Zhou, the CTO, both of whom are former Meta employees.

The startup has raised $16 million in total funding to scale its infrastructure. The capital includes:

• An $8 million seed round in July 2025, led by Susa Ventures.
• An $8 million funding extension.

Other investors in the company include Seedcamp, Fusion Fund, and Crane Venture Partners.

Memories.ai is developing a large visual memory model (LVMM), which the company launched in July 2025. According to Shen, the model could be compared to a smaller version of Gemini Embedding 2. To train the model, the startup uses LUCI (its internal data collection hardware), a wearable device worn by its data collectors to record video. Shen clarified that Memories.ai does not intend to become a hardware company or sell these data collection devices; rather, the company built LUCI because off-the-shelf video recorders did not meet its requirements. Additionally, Memories.ai has signed a partnership with Qualcomm to integrate its technology with Qualcomm processors starting later this year.

While the startup is already working with some wearable companies, its primary focus remains on building the underlying model and infrastructure. Shen indicated that the broader wearables and robotics market is probably not ready for commercialization yet. “In terms of commercialization, we are more focused on the model and the infrastructure, because ultimately we think the wearables and robotics market will come, but it’s probably just not now,” Shen said.

Why it matters

Memories.ai is attempting to bridge the gap between digital AI and physical world interaction by creating a visual memory layer. This technology serves as a critical infrastructure piece for future robotics and wearables, allowing physical AI systems to index and recall what they see.