How Ant Lingbo’s Full‑Stack Brain 2.0 Powers a ‘One‑Brain‑Multiple‑Robots’ Smart Pharmacy at WAIC

The 2026 World Artificial Intelligence Conference (WAIC) opened on July 17, featuring a "Top‑10 Treasure" showcase that highlighted Ant Group’s robot‑powered smart pharmacy built on the Ant Lingbo cross‑ontology embodied large model, alongside other high‑tech exhibits.

The smart pharmacy uses a "general brain" to coordinate robots of different brands and configurations, handling order receipt, intelligent sorting, and product delivery in a closed‑loop workflow. The solution has already been deployed in Guoda Pharmacy’s Shanghai store, and WAIC visitors can observe the "one‑brain‑multiple‑machines" capability in a real retail setting.

Industry analysts note that embodied‑intelligence is moving from focusing on the robot "body" to the "brain". Most existing robots still follow a one‑brain‑per‑robot pattern, requiring fresh data collection, training, and deployment whenever the hardware or task changes, which raises scaling costs.

To overcome this bottleneck, Ant Lingbo developed LingBot‑VLA 2.0, a universal intelligent base that was pre‑trained with 60,000 hours of high‑quality real‑world data. The model now supports more than 20 robot configurations from 17 vendors—including single‑arm, dual‑arm, bipedal, and wheeled platforms—and can control head, waist, end‑effector, and mobile chassis degrees of freedom.

LingBot‑VA 2.0 is the industry’s first native embodied world‑action model. Instead of fine‑tuning a digital‑world model, it was trained from scratch for physical‑world tasks, enabling robots to act, predict, and continuously adjust motions as the environment changes.

At the exhibition, several interactive demos highlighted the model’s capabilities: LingBot‑Depth 2.0 depth cameras fill missing spatial information in glass tanks and reflective surfaces; a handheld device lets users see LingBot‑Map generate real‑time 3‑D maps; and LingBot‑World 2.0 lets participants explore a generated world that evolves with their actions.

Ant Lingbo is currently conducting commercial pilots with partners such as Leju, Taihu, Guoda Pharmacy, and Longsheng in retail sorting, logistics, and industrial scenarios. By continuously feeding real‑operation data back into model training, the ecosystem of model, robot, data, and scene is converging toward a scalable, industry‑wide solution for "one brain, many machines".