Alibaba.com: When Global Sourcing Enters the AI Era

On September 6, 2024, Alibaba.com hosted another edition of its annual Co-Create conference in Las Vegas, drawing attention from businesses and tech enthusiasts worldwide. As always, Alibaba’s mission is to provide innovative tools to connect both buyers and sellers across the world and thereby boost the development of small businesses in the global market. Unlike previous years, this year’s event placed a clear and heightened emphasis on AI, reflecting the growing influence of AI across industries.

The retail e-commerce sector is not a new term. It has long been a fiercely competitive battleground with a huge number of players. On one hand, the barrier to entering e-commerce has significantly lowered, with easier access to information and tools, making it more accessible than ever. However, the influx of many new players has created a highly saturated “red ocean” of competition, making it harder for businesses to succeed. In this crowded market, where good and bad players are mixed together, and it’s difficult to distinguish truth from falsehood, small and medium-sized enterprises (SME) that are new to the industry face immense challenges. These obstacles make it tough for them to compete effectively.

The most eye-catching update introduced at this year’s Co-Create conference is the AI Sourcing Agent, presented by Kuo Zhang, President of Alibaba.com. The key innovation is in making the search process more intuitive, aligning it with how people naturally think. Traditionally, buyers had to rely on specific keywords, navigate multiple sources, and engage in lengthy back-and-forth communications with suppliers. They would manually summarize information on spreadsheets before making comparisons. However, with the power of Generative AI, this outdated approach is now being transformed, streamlining the entire process for greater efficiency and ease.

Recognizing that SME buyers on Alibaba.com come from diverse backgrounds and have varying levels of product knowledge, the AI Sourcing Agent offers two tailored user journeys. For those with limited market knowledge who want to gain deeper insights, the process begins with an “inspiration” phase. Buyers can initiate the search with simple prompts like, “I want to find potential bestsellers for Christmas.” The tool then identifies potential opportunities and presents a variety of solutions to choose from. After selecting a solution, the buyer is provided with multiple options, each accompanied by a summary of the competitive landscape, unit pricing, and supply-demand trends for further analysis.

Throughout the process, buyers can refine their prompts, allowing the tool to update and provide more detailed information, narrowing down options and recommending relevant suppliers based on contextual data and buyer behavior history. Comparisons can be made directly within the tool, eliminating the need for manual copy-pasting of information into spreadsheets, which can often lead to errors or missed details. This streamlined process significantly enhances efficiency and decision-making.

Additionally, the entire experience is conversational, much like interacting with an expert. This reduces communication gaps and costs, making the sourcing process both intuitive and time-saving, similar to the familiar AI tools buyers are accustomed to using.

For buyers who already have some basic knowledge about what they are looking for, the “sourcing” process follows a similar, but more targeted approach. For example, if a buyer asks a more specific question like “I am looking for a forklift for my farm,” the AI Sourcing Agent will generate a strategy that focuses on key factors and features they should consider. The tool will then provide summarized information to help buyers make decisions under different circumstances, allowing them to refine their search and find suitable suppliers. Traditionally, this process involves a lot of research during the browsing phase—constantly searching for similar products or visiting different stores to compare. However, we’ve now combined these steps into a more streamlined process. We’ve integrated research, selection, and comparison into one tool. In the past, you had to actively search for the product you wanted, manually select a range of options, and then summarize your findings across different suppliers and products. Now, we’ve consolidated these tasks. For example, when looking for similar products, not all of them are worth comparing. Only the ones that meet certain criteria deserve further attention, and our tool helps identify those key products for comparison.

Even if buyers have a clear idea of what they want, this process remains flexible. Extensive pre-matching of specifications is done automatically, streamlining the search. Should users want to adjust any details during sourcing, they can easily modify their preferences. Alibaba.com has also added a feature to tackle traditional sourcing challenges—where buyers, despite having a few preferred options, still want to explore additional possibilities and compare prices. This helps buyers feel more confident in their decision-making by thoroughly evaluating all available options. Afterall, from a business perspective, the more suppliers you have, the more likely you are to secure lower prices by negotiating or finding more competitive sellers.

The key reason a conversational and one-stop search process is now possible lies in the use of generative AI, which goes beyond traditional search engines that extract information solely from structured data. With generative AI, Alibaba.com can now incorporate previously difficult-to-process data, such as reviews and user-generated content (UGC), which were once challenging to analyze and interpret. The advanced models have a strong ability to understand and summarize semantics, enabling us to uncover valuable insights hidden within UGC—insights that often reveal emerging trends and consumer signals. This capability allows us to detect market trends earlier and identify consumer pain points more effectively, particularly through the rich data found in UGC.

The mission sounds attractive; however, concerns follow closely behind. For example, for small sellers on Alibaba.com, without paid ads to boost their rankings, will they be overlooked in the vast sea of information by AI and receive less attention compared to larger players? Indeed,in traditional e-commerce, businesses often pay to boost their rankings, but with AI-driven selection, this process is evolving. Instead of relying on paid promotions, AI uses a more organic approach by analyzing a wide range of data, not just from internal sources like Alibaba‘s international platform, but also from across the web. This means that even if you’re not listed on the international platform, your products can still be included in comparisons. This represents a more sustainable business model, where selection is based on the actual selling points of the products rather than on bidding for visibility.

Also, my personal experience tells me AI tools are not always reliable, how could I trust the information it provides? Will it have hallucinations? Currently, Alibaba.com is using Retrieval-Augmented Generation (RAG) to minimize such risks. where essential knowledge is pulled from a knowledge base to guide the model’s responses. This approach reduces the chances of hallucinations by ensuring that the information generated is based on reliable sources. Additionally, if the AI encounters something it doesn’t know, it is programmed to acknowledge that, rather than making guesses that could lead to factual inaccuracies.

Furthermore, the guidance provided is always reviewed by industry experts to ensure accuracy. After this initial phase, the process flows similarly to previous versions, especially in managing post-inquiry communications. For instance, after a buyer sends an inquiry, there is often extensive communication between the buyer and multiple sellers. Traditionally, buyers had to manually sift through individual chat histories. Now, the AI souring tool helps organize these conversations, managing the follow-up more efficiently, and also verifies information by cross-checking the seller’s statements with historical data, ensuring consistency. This verification is part of the AI’s summary, helping the buyer manage information more effectively.

Certainly, supply chain management is not a collection of isolated actions but an interconnected ecosystem, with sourcing being just one part of the whole. Beyond sourcing, Alibaba.com has also introduced new financial support options. In collaboration with Mastercard and powered by Cardless, the platform now offers consumers a choice between 3% cashback or 60-day interest-free payment terms, with a spending limit of up to $40,000 per year. After reaching that limit, cardholders can continue to earn 1% cashback on purchases made on the platform.

Currently, the AI Sourcing Agent focuses primarily on buyers, helping to test the product and gather feedback. However, Alibaba has plans to eventually integrate both the buyer and seller sides into a single tool, providing a more comprehensive view of the dynamics in this two-sided marketplace.

Starting a small business is never easy, and the development of AI is not meant to replace the value of interpersonal connections. Instead, it aims to reduce barriers between people, enhance communication, and allow entrepreneurs to focus on what truly matters. As the era of AI unfolds, the question remains: how will it continue to reshape industries? Alibaba.com is still in the process of exploring these possibilities, and we look forward to what the future holds.