Ant Brain Full-Stack 2.0 Debuts at WAIC, Powering Smart Pharmacy With Unified Intelligence

The 2026 World Artificial Intelligence Conference (WAIC) kicks off on July 17, with the prestigious "Treasures of the Exhibition" awards taking center stage. This year’s top ten selections highlight groundbreaking innovations, including Ant Group’s "Robotic Intelligent Pharmacy Powered by Ant Lingbo’s Cross-Entity Embodied Large Model" and Sustech’s "Shenwu 8000 - National-Production Ten-Thousand-Card AI Super Cluster," alongside other cutting-edge technological breakthroughs.

Selected by the conference organizing committee for their technical depth, market potential, scalability, and social impact, the "Treasures of the Exhibition" honor projects that demonstrate exceptional industry contributions and technological leadership. With hundreds of top-tier products submitted annually, only ten are chosen, making this award a highly coveted recognition in the tech community.

The award-winning intelligent pharmacy, co-developed by Ant Group’s Ant Lingbo and Guoda Pharmacy, leverages a "universal brain" to coordinate robots from various brands and configurations. This system automates order processing, intelligent sorting, and product delivery—a solution already deployed in Guoda Pharmacy’s Shanghai store. WAIC attendees can witness Ant Lingbo’s "one brain, multiple machines" capability in action within real-world retail environments.

From Order Receipt to Delivery: One "Brain" Orchestrates Diverse Robots

At the Ant Lingbo booth (H3-B302, Shanghai World Expo Exhibition Center), three robots—Lepu, Xinghaitu, and Ant Lingbo’s proprietary R-2—are powered by the embodied base model LingBot-VLA2.0. These robots autonomously assign tasks based on random orders, seamlessly handling order receiving, picking, and delivery. They identify medications, map locations, understand spatial relationships between shelves, people, and equipment, plan continuous tasks, and translate decisions into actions across different robotic bodies.

This solution is already operational in Guoda Pharmacy’s Shanghai store, requiring no special space modifications. The robots integrate smoothly into existing workflows, coexisting with staff and customers. During night shifts, they assist pharmacists in sorting medications, reducing repetitive labor and enhancing service efficiency for 24-hour pharmacies.

As robot hardware, components, and motion control advance rapidly, the embodied AI industry is shifting focus from "body" to "brain." Many robots still operate in a "one machine, one brain" mode, requiring data collection, training, and deployment for each new body, task, or scenario—increasing costs for large-scale implementation.

To overcome this bottleneck, Ant Lingbo is building a universal intelligent foundation for the industry. LingBot-VLA2.0 integrates 60,000 hours of high-quality real-world data during pre-training and supports over 20 robot configurations from 17 manufacturers, including Lepu, Zhiyuan, Yushu, and Galaxy General. It covers single-arm, dual-arm, bipedal, and wheeled forms, supporting full-body degrees of freedom such as head, waist, end effector, and mobile base.

Full-Stack Brain 2.0 Unveiled: Enhanced Vision, Clearer Thinking, Greater Efficiency

At WAIC 2026, Ant Lingbo showcased "Full-Stack Brain 2.0," demonstrating how a universal brain enables robots to "see more clearly, think more clearly, and work more efficiently." This model spans the entire stack of spatial intelligence, dexterous operations, and environmental feedback.

Notably, LingBot-VA2.0 is the industry’s first embodied-native world action model. Rather than fine-tuning digital world models, it is pre-trained from scratch to meet the demands of physical-world tasks, enabling robots to act and predict simultaneously while rapidly adjusting to environmental changes.

Beyond the intelligent pharmacy, several interactive models are on display. Visitors can observe how LingBot-Depth2.0’s depth camera reconstructs missing spatial information for fish tanks, water, and reflective objects; move a handheld device to see real-time 3D map generation by LingBot-Map; and use a controller to explore LingBot-World2.0, experiencing how the generated world evolves continuously with user interactions.