Many AI tools can see a video today and summarize what is happening, but things get a little difficult when you ask modeling questions about multiple videos and videos that cover many hours.
This is a great limit for security companies who want to use AI to clean thousands of hours from different cameras, as well as marketing companies that want to study different video and product shoots.
Memories.Ai He wants to deal with this problem with the AI platform that can edit up to 10 million hours of video. For companies with lots of videos to analyze, the start wants to provide a context of context, complete with searching indexing, labels, sections and agglomeration.
His co -founder, Dr. Shawn Shen, was a researcher at Meta’s reality workshops, followed by his doctorate and his counterpart ENmin (Ben) Zhou worked at META as a mechanical learning engineer.
“All the top AI companies, such as Google, Openai and Meta, focus on producing end -to -end models. These features are good, but these models often have limitations around understanding the video frame more than one or two hours,” Shen told Techcrunch.
“But when people use visual memory, we are sifting a large framework of data. We were inspired by it and wanted to build a solution to better understand in many hours better,” he said.
To this end, the company has now raised $ 8 million in a seed funding round led by Susa Ventures and with the participation of Samsung Next, Fusion Fund, Crane Ventures, Seedcamp and Creature Ventures. Shen said the company had initially aimed to raise $ 4 million, but ended up with an excessive round due to investors’ interests.
TechCrunch event
Francisco
|
27-29 October 2025
“Shen is an extremely technical founder and is obsessed with pushing the boundaries of understanding and intelligence,” said Misha Gordon-Rowe, a partner of Susa Ventures. “Memories.Ai can unlock a lot of first -part visual intelligence data. We thought there was a gap in the market for long visual intelligence that attracted us to invest in the company,” he added.
Samsung then had a slightly different dissertation: Samsung’s investment arm sees the memories. Its solution is useful for consumers.
“One thing we liked for memories. It is that it could make a lot of computers on devices, it means that you don’t necessarily need to store video data in the cloud, this can unlock better security applications for people who are worried about the placement of security cameras at home due to worries about the protection of private life.”
Memories.Ai says he uses his own technological stack and models to perform analyzes. First, it removes the noise from the videos and passes the exit through a compression layer to store only significant data. Then there is a layer of indexing, which makes the video that can be searched for data (using natural language questions) with fragmentation and labels. There is also a aggregation layer that summarizes data from the index, helping to create reports.
Currently, the start serves two types of companies: marketing and security. Marketing companies can use the boot tools to look for trends related to their brands in social media and identify the kind of video they want to make. Memories.Ai also provides tools for traders to create these videos.
The company is also working with security companies to help them analyze security plans to determine potentially dangerous actions from people in the video with motifs.
Currently, companies working with memories. They have to upload their library to the platform to analyze the clips. But Shen said that in the future, his clients would be able to create a more common content and synchronization content more easily. The plan is to enable customers to ask questions such as: “Tell me all about the people I have interviewed last week.”
Shen provides an assistant AI who can win the frame in a user’s life through his photos or when activating smart glasses. It also sees technology playing a role in the training of humanoid robots to do complex tasks or help self-guiding cars remember different paths.
The company currently employs 15 people and plans to use the fund to increase its team and improve its search.
Memories.ai rises against similar newly established businesses, such as MEM0 And Letta, who are working to provide a Memory Memory for AI models, although they offer limited video support right now. It must also face companies such as Twelvelabs and Google, who work to help AI models understand the videos.
Shen, however, feels that his company’s solution is more horizontal, which would allow her to work with different video models.
