South Korea’s high cell provider SK Telecom Co. mentioned on Friday it’ll make investments $3 million in Twelve Labs Inc., a Korean AI-powered video analysis startup, to make use of the startup’s expertise in its AI agent companies.
Established in San Francisco in 2021, Twelve Labs develops AI-based multimodal video understanding and search applied sciences.
The startup is thought for attracting $50 million in a Series A funding round led by international chip designer Nvidia Corp. in June.
Nvidia’s enterprise capital affiliate NVentures and New Enterprise Associates, a brand new investor in Twelve Labs, collectively led the Sequence A spherical. Present international buyers together with Index Ventures, Radical Ventures and WndrCo, led by DreamWorks co-founder Jeffrey Katzenberg, and Seoul-based Korea Funding Companions additionally joined the spherical.
The prevailing buyers participated in pre-Series A funding of about $10 million final October, in which Nvidia made its first funding in a Korean generative AI startup.
SK TELECOM’S AI AGENT
By the funding, SKT expects to boost its AI Agent, an AI butler service, by combining the 2 corporations’ AI experience.
The 2 corporations additionally agreed to collaborate on creating and advancing applied sciences for implementing multimodal AI in safety and public security functions, comparable to AI surveillance programs.
Not like conventional surveillance programs the place a single operator needed to monitor quite a few CCTV feeds for lengthy hours, Twelve Labs’ multimodal AI mannequin permits for fast looking and summarizing of key incidents, actions and people.
The 2 corporations additionally plan to collaborate on creating and advancing applied sciences to use multimodal AI in areas like safety and public security, together with AI monitoring programs.
SK Telecom mentioned Twelve Labs will be part of the Okay-AI Alliance, a bunch of Korean corporations selling AI expertise, to collaborate with different members in fostering Korea’s AI ecosystem.
“By the partnership with SKT, we sit up for offering our video basis fashions to numerous business use instances and supply actual worth in day by day workflows in the ecosystem,” mentioned Jae Lee, CEO of Twelve Labs.
Lee Jae-shin, head of AI Progress Technique at SK Telecom, mentioned: “By the cooperation of the 2 corporations, we’ll additional strengthen our competitiveness in the multimodal AI subject.”
TWELVE LABS’ AI TECH
Multimodal AI is used for a machine studying mannequin, in which numerous information varieties together with picture, textual content, speech and quantity are mixed with intelligence processing algorithms for correct and complex outputs.
Based mostly on the multimodal mannequin, Twelve Labs analyzes photos and sounds in a video and matches them to human language. The mannequin can also create textual content based mostly on the video content material, edit a short-form video and categorize movies by a sure commonplace.
The expertise boosts effectivity in creating YouTube Shorts, establishing commercial methods for movies and even discovering lacking individuals by analyzing closed-circuit tv (CCTV) footage, in keeping with the startup.
Twelve Labs has built-in a few of Nvidia’s framework and companies inside its platform, together with the NVIDIA H100 Tensor Core Graphic Processing Unit and NVIDIA L40S GPU, to enhance its video understanding expertise.
In March, Twelve Labs launched the multimodal mannequin Marengo-2.6, which allows numerous video, textual content, picture and audio search duties and in addition launched a beta model of Pegasus-1, which is particularly designed to grasp and articulate video content material.
Write to Seung-Woo Lee at leeswoo@hankyung.com
In-Soo Nam edited this text.