EXCLUSIVE: Luma launches creative AI agents powered by its new ‘Unified Intelligence’ models

Luma, an innovative startup specializing in AI-driven video generation, has officially introduced Luma Agents, a cutting-edge solution designed to streamline creative processes across various media formats, including text, images, videos, and audio. These agents leverage Luma’s advanced Unified Intelligence models, which utilize a sophisticated multimodal reasoning system. Targeted primarily at advertising agencies, design studios, and corporate marketing teams, Luma Agents promise to transform traditional workflows. According to Amit Jain, the CEO and co-founder of Luma, these agents can autonomously generate and plan creative content while also integrating with other AI models, such as Google’s Veo 3, ByteDance’s Seedream, and ElevenLabs’s voice capabilities. The foundation of Luma Agents is the Uni-1 model, the first in its family of Unified Intelligence models, which has been meticulously trained in multiple domains including audio, video, and spatial reasoning. Jain described this model as capable of “thinking in language and visualizing in pixels,” a concept he refers to as “intelligence in pixels.” Future iterations of the model are expected to enhance audio and video outputs further. Jain emphasized that Luma Agents are not just tools; they are a transformative approach to how creative work is executed. With existing partnerships that include notable global ad agencies and brands such as Adidas and Mazda, Luma is already demonstrating the capabilities of its new platform. One standout feature of Luma Agents is their ability to maintain continuous context throughout the creative process, allowing for refined outputs through iterative self-critique. This capability, Jain explained, is crucial for enhancing the quality of results and resembles the iterative processes used in coding. In contrast to the often cumbersome workflows currently seen in the creative industry, Luma Agents provide a more intuitive experience. Jain noted that users can interact with the system in a conversational manner, steering the creative direction without the need for repetitive prompting. To illustrate Luma Agents’ potential, Jain showcased how a simple 200-word brief alongside an image could yield numerous creative ad concepts within a short timeframe. In a striking example, the agents were able to transform a $15 million ad campaign into localized versions for multiple countries in just 40 hours, all while adhering to quality standards. Currently accessible via API, Luma Agents will be rolled out gradually to ensure users experience seamless integration without workflow interruptions, marking a significant leap in the creative capabilities of AI technology.