Luma Launches AI ‘Agents’ for End-to-End Creative Work

Luma Launches AI ‘Agents’ for End-to-End Creative Work

Luma has introduced Luma Agents, a new category of artificial intelligence tools designed to execute end-to-end creative work across text, image, video, and audio. The company said the system is built on its newly developed “Unified Intelligence” architecture and is already being deployed with enterprise partners including Publicis Groupe and Serviceplan Group.

According to Luma, the agents are designed to help agencies, marketing teams, studios, and enterprises manage creative workflows from initial briefing to final production. The system maintains shared context throughout the process and can coordinate different models, tools, and iterations within a single environment.

“Creative teams shouldn’t have to spend their time orchestrating tools,” said Amit Jain, co-founder and chief executive officer of Luma. “Agents maintain context, coordinate execution, and advance projects so teams can focus on direction and strategy.”

Luma said the agents allow teams to manage projects from planning through delivery, work with text, image, video, and audio in the same workflow, and develop multiple creative directions simultaneously. The platform can also integrate with enterprise tools and production systems through application programming interfaces (APIs).

Serviceplan Group has integrated the system into its creative workflows as part of its internal AI ecosystem. Alexander Schill, global chief creative officer at Serviceplan Group, said the integration allows teams across more than 20 countries to collaborate more efficiently while delivering creative output more quickly.

The agents are powered by Luma’s Unified Intelligence architecture, which the company says differs from traditional AI systems that rely on separate models for language, images, and video. Instead, Luma trains a single multimodal reasoning system capable of processing and generating multiple formats within the same architecture.

The first model built on this system, called Uni-1, combines language and image tokens within a shared framework to support reasoning and content generation in a single process.

Luma said the platform includes enterprise safeguards such as intellectual property ownership for customers, automated content review, documentation of human involvement, and mandatory human review workflows before public release.