Finding a supplier can leave a small business with dozens of conversations and a spreadsheet full of unfinished work.
Some suppliers reply quickly. Others take days. Prices arrive alongside different minimum orders, certifications, delivery arrangements, and payment terms. Someone has to gather the answers, make them comparable, and keep the conversations moving.
Alibaba.com, a global platform connecting business buyers and suppliers, built Accio Work as a workspace where AI agents help businesses carry out tasks like this. Its president, Kuo Zhang, says the supplier matching and comparison work can now happen within a day (versus weeks or months before), and sometimes even within hours.
In Episode 244 of The Artificial Intelligence Show, part of our AI Transformations series, Zhang walks through how the process works. His example offers a useful blueprint for a developing approach to global sourcing: Give an agent the buyer's requirements and budget, then connect it to the information and tools needed to act on them.
How Global Sourcing Works Without AI
Zhang describes how small businesses typically source a product overseas.
The buyer first writes down what they need, then sends that request to multiple suppliers. Replies arrive at different times. The buyer transfers information into a spreadsheet to compare price, minimum order quantity, certifications, quality, weight, and other requirements.
Once enough information is available, the buyer narrows the list and starts another round of conversations. This time, the questions concern the details that could make or break the order: things like production requirements, shipping arrangements, delivery, and payment.
Zhang says the process of finding a supplier and working through the order details can take weeks or months. Across the broader journey from an idea to a delivered product, he describes roughly 60 steps, complicated by different time zones, languages, and cultures.
Collecting all the upfront information is a large part of the job. The buyer has to explain the request, wait, follow up, and understand what each supplier is offering before a useful comparison exists.
Those repeated exchanges are central to the work Alibaba.com is now giving agents.
Start With the Buyer's Requirements and Budget
Accio Work is the company’s agentic AI platform for buyers. Using it, the buyer begins by explaining what they need and what they can spend.
For an inexperienced buyer, even that can be difficult. Zhang says some business owners do not know how to prepare a design pack that tells a manufacturer what to make. Accio can help turn a product idea into that more useful description.
The agent searches for suppliers that match those requirements, helps communicate with them, and organizes the information into the fields the buyer needs to compare. The customer is no longer responsible for carrying every piece of information from a conversation into the comparison.
Zhang describes the time needed to reach that stage as:
“Probably just within a day; if you are lucky, in hours.”
That is the supplier research and comparison stage. Further discussions, a contract, and fulfillment follow. The agent can continue helping with communication as the buyer works through order quantities, pricing, shipping, and payment terms.
This changes where the buyer needs to concentrate their attention. Instead of spending so much time gathering replies, the buyer has an organized set of options from which to continue the commercial decision.
It also explains why the quality of the initial request matters. The agent is matching suppliers against the needs and budget the buyer gives it. Clear requirements give that work a useful direction.
The Agent Needs Tools and Business Context to Act
Understanding a buyer's request does not by itself give an agent a way to contact suppliers or use their replies.
Zhang describes three parts of the Accio Work system. The model interprets the request. The harness is the surrounding software, including the tools, logic, and guardrails that let it act. Context supplies the business information, including the buyer's requirements, supplier conversations, and platform data.
The conversations matter because a sourcing decision develops as suppliers respond. The information needed to compare an offer may be different from what was available when the buyer first described the product. Useful assistance depends on keeping the work connected to that developing information.
This has required work on the platform itself. Zhang says Alibaba.com's engineers are making its services easier for agents to use. A system designed for a person clicking through a website also has to support the software now also carrying out the work.
The Business Owner Still Supplies the Purpose
A clearer request and an organized supplier comparison can reduce work for the owner. They do not decide what business the owner should build.
Zhang brings the focus back to the questions that give the sourcing work its purpose:
“What problem you are going to tackle. Who are your customers.”
Leaders evaluating a similar workflow need to make those responsibilities explicit. Define the outcome and constraints a person needs to supply. Identify the information the agent needs and the systems it must use. Then decide which results and commitments require a person's review before the process continues.
A supplier comparison, for example, is a different kind of output from a commitment to spend money. Designing around those differences helps a business decide how far to delegate each part of the work.
Before automating the spreadsheet, look at the conversations someone had to manage to fill it.
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.
Mike Kaput
Mike Kaput is the Chief Content Officer at SmarterX and a leading voice on the application of AI in business. He is the co-author of Marketing Artificial Intelligence and co-host of The Artificial Intelligence Show podcast.

