Memories.ai is a fast-growing AI startup focusing on advanced video understanding, and has managed to shut down an over-seeded $8 million seed funding. The investment, led by Sousa Ventures, with significant participation from Samsung Next, Fusion Fund, Crane Ventures, Seed Camp and Creator Ventures, will promote the company's ambitious plan to process and contextualize thousands of hours of video footage, addressing key gaps in its current AI capabilities.
While many existing AI tools can summarise short video clips, imposing multiple videos or video analysis over a wide range of time will significantly reduce its effectiveness. This limitation poses a major challenge for industries like security. Security requires cleaning up a large amount of camera footage and marketing.
Memories.ai aims to overcome this hurdle with its own AI platform designed to process an astounding 10 million hours of video. The startup's core products provide a “context layer” for businesses with large video libraries, allowing searchable indexing, intelligent tagging, accurate segmentation and robust aggregation of video data.
The company's co-founder, Dr. Shaun Shen, brought a strong research background from Meta's reality lab and was a research scientist while earning his PhD. His co-founder, Enmin (Ben) Zhou, also has important experience as a machine learning engineer at Meta.
“All top AI companies, such as Google, Openai, and Meta, focus on producing end-to-end models. These features are great, but these models often have limitations on understanding video contexts over an hour or two,” Shen explained to TechCrunch. “However, when humans use visual memory, we sifted through the large context of data. We were inspired by this and wanted to build solutions for hours to better understand videos.”
The interest of strong investors, which has grown from the round's initial target of $4 million to $8 million, highlights the recognised market needs. Misha Gordon-Rowe, partner at Susa Ventures, praised Shen as “a highly technical founder” with a deep commitment to “pushing the boundaries of video understanding and intelligence.” Gordon-Rowe highlighted that Memories.ai can unlock many first-party visual intelligence data with its solutions. We felt there was a gap in the market for long contextual visual intelligence.
The Samsung Next investment paper provided a clear perspective and highlighted the technology's potential in Memories.AI's consumer applications. “One of the things we like about memories is that we can do a lot of on-device computing, meaning we don't have to store video data in the cloud. The processing capabilities on this device can greatly improve end-users' privacy and data security.
Memories.ai employs a sophisticated in-house technology stack and models for analysis. This process removes noise from the video, followed by a compression layer that holds only important data. The index layer makes this video data searchable using natural language queries with segmentation and tags. Finally, the aggregation layer summarises indexed data to facilitate the creation of comprehensive reports.
Today, Memories.ai serves two major client segments: marketing and security. Marketing companies leverage the platform to identify brand trends across social media videos and notify you of the creation of new video content. To security companies, startups provide tools to analyze surveillance footage and infer through patterns to detect potentially dangerous actions.
Currently, clients are uploading video libraries for analysis, but Shen envisions future integrations that allow customers to easily sync content from shared drives. He foresees a future in which users can ask AI assistant contextual questions like, “Tell me all about the people I interviewed last week.” Looking further, Shen believes that the technology can play a transformative role in training humanoid robots for complex tasks and assisting self-driving vehicles that remember diverse routes.
Memories.ai currently has a team of 15 employees and plans to leverage new funds to expand the team and enhance search capabilities. Startups are navigating the competitive landscape and are facing other “memory layer” AI startups such as MEM0 and Letta. They also compete with established players such as Twelvelabs and Google, both actively developing AI models for video understanding.
However, Shen is confident in her memory. The AI position argues that its solution is more “horizontal” and allows for seamless integration with a variety of existing video models.
