Alibaba.com built its business connecting buyers and suppliers around the world. Now it is building AI agents to help those customers do more of the work of commerce.
For an entrepreneur or small business, Alibaba.com is a place to find manufacturers and source products. But finding a supplier is only part of the job. A business still has to explain what it needs, compare offers, negotiate terms, and prepare products for sale.
Accio Work, the company's agentic AI workspace, is designed to help with that broader work. Its agents assist with tasks that range from researching a product opportunity and communicating with suppliers to managing listings across ecommerce platforms.
In Episode 244 of The Artificial Intelligence Show, part of our AI Transformations series, Kuo Zhang, president of Alibaba.com, explains how the company got here and what had to change along the way.
The story began with an AI product search tool. It grew into a broader agent business, then reached back into Alibaba.com's established platform and the way its teams build and release software.
Before Accio Work, there was Accio, an Alibaba AI search product that helped buyers find products and suppliers.
Traditional product search depended on entering keywords. Zhang says buyers increasingly wanted to describe their needs in sentences, images, and documents. AI gave the company a way to understand those more detailed requests.
Once Accio launched, customers used it for more than finding an existing product. They explored ideas, investigated what was already on the market, and identified opportunities to improve on it. They then used AI to turn those ideas into a design pack: the information a supplier needed to understand what they wanted made.
That solved a problem a search result alone could not. Buyers sometimes struggled to explain their ideas to suppliers. Zhang says the AI-generated design packs made buyers appear more professional and reduced back-and-forth discussions and disputes over their requirements.
To help further, Accio expanded into more of the sourcing process, including helping buyers negotiate. The product was moving from helping a customer find something to helping them carry out the steps needed to buy or make it.
The same customers also needed help after they sourced a product. Many sold through several platforms, each with its own information and tasks to manage.
A seller might have storefronts on Shopify, Amazon, and eBay. Updating a product could mean repeating work across platforms, adapting images and listings, and following different rules. Understanding the business could require pulling information into separate spreadsheets.
Zhang describes the underlying customer request simply:
“They want to manage the business as one.”
Alibaba.com expanded into this work through Accio Work, a broader agent workspace built on its earlier search and sourcing capabilities.
For a seller preparing a new product, the work might begin with market research, continue into finding a manufacturer, and then move into listing the product across storefronts. Zhang describes Accio Work providing agent assistance across that sequence, including research into competitors and potential margins.
That gave Alibaba.com a broader role in helping customers operate their businesses, including work on platforms beyond its own marketplace.
Building that agent business brought another challenge: It had to develop quickly and still connect to an established platform with existing customers and services.
Through all of this, Alibaba.com had an established business to maintain. Its platform needed new features to work with existing ones, and customers needed continuity. Zhang describes a release process that had typically operated on a biweekly or monthly schedule, with testing before changes went live.
Accio, however, operated as a startup team within Alibaba Group. It released a new version of the product daily thanks to AI-assisted code.
For Zhang, the difference became a morning routine.
“Every morning that I get up, the first thing is that I upgrade this software or this agent to see what's new there.”
Giving the new team its own pace was one part of the change. Making its work useful across the existing business required another.
Alibaba.com's engineers worked on making the existing platform's services accessible to agents. Those services had to support software carrying out tasks as well as people using the website. Without that connection, the new agent product would have had a limited ability to act through the company's established platform.
The company then brought Accio into Alibaba.com through a feature called AI mode. It gives customers using the marketplace access to Accio's help with work such as researching requirements and finding suppliers. Zhang says Alibaba.com now releases updates daily as well.
Zhang connects the faster updates to Accio's integration into the core platform. The startup team's pace, in his account, did not remain confined to a separate product.
Today, Accio Work helps with activities that span the buyer's business. Zhang's larger ambition is for agents representing buyers and suppliers to work directly with one another.
He calls that direction agent-to-agent commerce.
His reasoning starts with a human constraint. A marketplace can contain an enormous supplier network, but a buyer has time to inspect only some listings and speak with a limited number of companies. The network's size does not automatically make all of it usable.
Zhang believes agents could compare needs and capabilities across more of that network, finding matches based on a buyer's requirements, budget, and timing. That would expand the platform's role in helping businesses conduct trade.
That is a vision Alibaba.com is pursuing. Getting there will require continued learning, and Zhang is candid about how quickly the company's understanding has changed.
“In the AI era, three years means forever.”
He says knowledge gained over the previous six months would have changed earlier decisions. The steady point in that uncertainty was the customer problem: Small businesses needed help with complicated, repetitive work that limited their ability to trade.
Alibaba.com's transformation involved both the work it could help customers do and the way it developed and delivered that help. Its experience points to several decisions for leaders building agents into an established business:
For leaders pursuing a similar shift, the questions are connected: What work could agents help customers carry out, and what must change inside the company to make that possible? Alibaba.com had to work on both.
Did you enjoy this transformation story? Go deeper on how AI is reshaping work and business with The Artificial Intelligence Show. Each week, we break down what matters in AI and what leaders should do about it.