CJ Logistics and physical AI firm RealWorld signed a memorandum of understanding to jointly develop a robotics foundation model (RFM) for logistics automation, the companies said on the 21st. The deal sets up a proof-of-concept at actual logistics sites, with plans to scale into logistics-as-a-service (LaaS) and joint overseas expansion - a move that puts task-specific AI directly into warehouse picking, packing, and delivery workflows.
What the partnership covers
The two companies will combine RealWorld's RFM technology with CJ Logistics' operating infrastructure to build solutions for real logistics processes. RealWorld's model, RealDex One (RLDX-1), uses visual and sensor data to understand its surroundings and determine the actions needed for a given task. CJ Logistics has already been deploying AI humanoid robots at its sites, and the company said learning-based automation is necessary to handle the wide range of products and working environments it manages.
They will jointly select specific logistics processes for the proof-of-concept and verify core movements. After that, they plan to gradually expand the range of processes and raise the technology's maturity. The end goal is commercialization of LaaS - packaging order processing, storage, picking, packing, delivery, returns, and data analysis as a service.
Technology stack and rationale
RealDex One falls into the category of physical AI - models that perceive environments and execute physical tasks rather than just generating text or images. CJ Logistics' existing automation efforts made the partnership a logical step. The company needed a logistics-specific model that could adapt to varied SKUs and workflows, and RealWorld holds RFM technology purpose-built for that problem.
"By combining CJ Logistics' logistics sites and operating capabilities with RealWorld's RFM and physical AI technology, we will solve problems at actual logistics sites and build a logistics-specific physical AI model that can be expanded to the global market," said RealWorld CEO Ryu Jung-hee.
Global expansion plans
The agreement also targets overseas markets. The two companies intend to link RealWorld's global partner network with CJ Logistics' commercialization capabilities. They will pursue joint market entry and strategic cooperation, and explore global LaaS opportunities using the logistics-specific RFM verified in Korea as the foundation.
Why this matters for operations and product development leaders
This partnership signals a shift from general-purpose robotics toward task-specific AI models trained on real warehouse environments. For operations and supply chain managers, the LaaS model could lower the barrier to automation by turning capital-intensive robotics into an operational expense. Product development and R&D teams should watch how the RFM handles variability - the model's ability to generalize across products and layouts will determine whether this stays a pilot or becomes a template for logistics automation at scale. Professionals evaluating similar deployments may want to build internal benchmarks around picking accuracy, changeover time, and integration with existing warehouse management systems.
Your membership also unlocks:
