Alibaba puts agentic AI to work in e-commerce — at massive scale

• Alibaba rolls out autonomous digital workers to millions of Taobao and Tmall merchants.
• It bets the token economy, not the chatbot, is where AI value lands.

The conversation surrounding AI agents has been running ahead of the evidence for some time now. Most enterprise deployments are still in pilot mode, confined to controlled workflows, narrow tasks, and limited autonomy. What Alibaba is doing on Taobao and Tmall this month is something different: it is putting agentic AI e-commerce into live commercial operation, at a scale that most organisations haven’t reached in any single AI programme.

As per a recent South China Morning Post reportthe company unveiled the service at Tmall’s annual TopTalk merchant summit in Shanghai last week, announcing that agentic AI abilities would be extended to millions of merchants on both platforms before the end of March.

Built on its existing merchant tool Business Advisor, the service gives sellers a 24/7 autonomous layer – one that can handle customer queries, push vouchers and adjust product pricing in real time without human instruction.

Xu Haipeng, who oversees merchant platforms at Taobao and Tmall, framed the change plainly at the event. Autonomous, execution-oriented AI had moved from aspiration to operational reality, he said, and over the next one to two years, the standard operating model of e-commerce would evolve into a collaboration between human and digital employees.

That last phrase is worth sitting with. Not AI-assisted employees. Not AI-augmented workflows. Digital employees – a distinction that signals a meaningful change in how Alibaba is thinking about what these systems actually are.

The infrastructure behind the announcement

On March 16, Alibaba created the Alibaba Token Hub business group, consolidating its Tongyi Laboratory, Qwen AI unit, and Wukong enterprise platform under CEO Eddie Wu – a restructure designed specifically around what the company calls the agentic era.

Wukong, Alibaba’s AI-native enterprise platform, was unveiled a day later as the product of that new group, capable of coordinating multiple agents to handle complex tasks in a single interface.

The Taobao and Tmall merchant deployment represent Alibaba’s argument that agentic AI e-commerce requires the consumer-facing layer and the enterprise infrastructure to be moving in the same direction.

To power that consumption, Taobao and Tmall said they plan to inject one trillion tokens into their AI services, targeting customer service scenarios specifically, upgrading their existing AI tools from reactive chatbots into proactive shopping guides capable of handling complex tasks.

What the numbers say

Over the past year, AI tools provided by Taobao and Tmall have been used by five million merchants, helping them save an estimated 100 billion yuan in costs. The AI customer service tool Dianxiaomi has served 300 million customers while raising transaction conversion rates by 30%.

China’s AI agent sector is projected to scale from under US$1 billion in 2024 to over US$30 billion by 2028, with the enterprise segment attracting the bulk of structural investment. Gartner estimates that 40% of enterprise applications will adopt task-specific AI agents by 2026, up from less than 5% in 2025.

The structural advantage that matters

What separates Alibaba’s agentic AI e-commerce deployment from most Western enterprise counterparts is not the ambition. It is the plumbing. Alibaba’s models can theoretically guide a consumer to a product, complete the purchase, process payment and arrange delivery – all in a single agent interface, without negotiating third-party APIs, because the data flows through systems Alibaba already owns.

The broader Alibaba ecosystem – Taobao, Tmall, 1688, Alipay and Alibaba Cloud – is being integrated into Wukong as modular agent skills, with third-party integrations. The agent does not need to leave the building to do its job.

Most enterprise AI deployments in markets outside China are still working through integration agreements, data access negotiations and fragmented vendor stacks. Alibaba is operating from a unified system it built and controls. That is not a marginal advantage.

What does this signal for enterprise AI teams

The Taobao and Tmall announcement is a proof point about where agentic AI deployment is headed and how quickly the gap between “exploring” and “operating” can close when the underlying infrastructure is already in place. Alibaba’s cloud president has said publicly that a single person equipped with AI agents could run a billion-dollar company in three to five years. The merchant deployment this month is the first meaningful evidence that the company is building toward that thesis not stating it.

Organisations still debating whether agentic AI belongs in their roadmap are now watching a competitor class that has moved past the roadmap entirely.

Want to experience the full spectrum of enterprise technology innovation? Join TechEx in Amsterdam, California, and London. Covering AI, Big Data, Cyber Security, IoT, Digital Transformation, Intelligent Automation, Edge Computing, and Data Centres, TechEx brings together global leaders to share real-world use cases and in-depth insights. Click here for more information.