Shawn Shen, co-founder of Memories.ai, posits that for artificial intelligence to truly thrive in the physical world, it must possess the capacity to retain visual information. His company, Memories.ai, is actively leveraging Nvidia AI tools to construct the foundational infrastructure that will enable wearables and robotics to both store and retrieve visual memories effectively.
At Nvidia's GTC conference on Monday, Memories.ai officially announced a strategic collaboration with the semiconductor titan. This partnership integrates Nvidia’s Cosmos Reason 2, a sophisticated reasoning vision language model, and Nvidia Metropolis, an application designed for video search and summarization, to further advance Memories.ai's cutting-edge visual memory technology.
Shen disclosed to TechCrunch that the genesis of Memories.ai stemmed from his and co-founder and CTO Ben Zhou’s work on the AI system powering Meta’s Ray-Ban smart glasses. This experience illuminated a critical gap: the practical utility of such AI glasses would be severely limited if users were unable to recall the video data they had recorded.
Their subsequent search for an existing visual memory solution for AI proved fruitless. Recognizing this unmet need, Shen and Zhou made the strategic decision to spin out from Meta and develop the necessary technology themselves.
“AI is already doing really well in the digital world, what about the physical world?” Shen remarked. He emphasized, “AI wearables, robotics need memories as well. … Ultimately, you need AI to have visual memories. We believe in that future.”
The concept of AI systems possessing memory is a relatively recent development. OpenAI, for instance, introduced the ability for ChatGPT to remember past conversations in 2024, refining this feature in 2025. Similarly, Elon Musk’s xAI and Google Gemini have also rolled out their own memory capabilities within the last two years.
However, Shen points out that these advancements have predominantly concentrated on text-based memory. While text-based memory is inherently more structured and easily indexed, it offers limited utility for physical AI applications that primarily interact with their environment through visual input.
Memories.ai, launched in 2024, has successfully secured $16 million in funding to date. This comprises an $8 million seed round completed in July 2025, followed by an additional $8 million extension. The funding round was spearheaded by Susa Ventures, with participation from notable investors including Seedcamp, Fusion Fund, and Crane Venture Partners, among others.
According to Shen, the successful development of this visual memory layer hinges on two key components: establishing the infrastructure required to embed and index video content into a retrievable data format, and simultaneously acquiring the extensive data necessary to train the model for these specific functions.
The company unveiled its Large Visual Memory Model (LVMM) in July 2025. Shen drew a comparison, noting that their LVMM could be considered a more compact version of Gemini Embedding 2, a multimodal indexing and retrieving model that was introduced earlier this month.
For the crucial task of data collection, Memories.ai engineered LUCI, a proprietary hardware device worn by their “data collectors” to record video for model training. Shen clarified that the company has no intentions of becoming a hardware manufacturer or selling these devices. Instead, they developed LUCI out of necessity, finding commercial off-the-shelf video recorders inadequate due to their focus on high-definition and battery-intensive video formats.
Further advancing its technology, the company has since released the second generation of its LVMM and forged a partnership with Qualcomm, enabling its model to run on Qualcomm’s processors beginning later this year.
Memories.ai is already collaborating with several prominent wearable companies, Shen confirmed, though he refrained from disclosing their identities. While current demand exists, Shen foresees even greater opportunities emerging in the wearables and robotics sectors in the future.
“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 articulated, outlining the company's strategic approach.
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.