Alibaba shifts focus from chatbots to robots and AI agents

Alibaba unveiled its first full suite of AI models for robots and agents on Tuesday. The models are designed to help machines identify objects, plan movements, call external tools and complete multi-stage tasks in real-world settings such as kitchens, warehouses and factory floors.

The models add capabilities beyond conversational chatbots. They include persistent memory, task planning, tool use and coordination between software and hardware so machines can run workflows with reduced human supervision.

DAMO Academy, Alibaba’s research unit, contributed earlier work including RynnBrain, an embodied AI model for physical reasoning, navigation and task planning that feeds into the new models.

Alibaba has also advanced related software and hardware. In May the company introduced Qwen3.7-Max, a large model built to handle long, complex tasks and reported a 35-hour autonomous run that made more than 1,000 tool calls. The company also revealed XuanTie C950, a 5-nanometre RISC-V processor intended for on-device and edge agent workloads where memory, coordination and repeated interactions matter.

The company presented the combination of chips, cloud infrastructure and foundation models as a way to deploy agents across commerce, logistics, cloud services and industrial customers. Alibaba positioned the technology for both industrial and commercial applications.

Alibaba chief executive Eddie Wu predicted there may one day be more agents and robots than people.

Marc Einstein, research director at Counterpoint Research, warned agents could ‘upend traditional Internet business models’ and cautioned that consequences could be severe for companies that are not prepared.

Other Chinese technology firms, including ByteDance, Zhipu AI and Baidu, are also developing agentic systems.

Demonstrations focused on object recognition, spatial understanding, motion planning and task execution in cluttered or changing environments. Alibaba presented use cases where robots perform multi-stage workflows such as booking logistics, managing supplier orders, updating enterprise systems or operating machinery with minimal human input.