The Heart of the Machine Original
Heart of the Machine Editorial Department
After a year and a half, generative AI's application-layer innovation has finally entered an explosive period.
Sora at the beginning of the year sparked a wave of AI + video generation. The ripples have not yet dispersed, and OpenAI's new SearchGPT has ignited the war of AI + search again.
AI + Search, this is actually what most people imagined about the application of generative AI technology when it first emerged: to change the traditional search engine. On this battlefield, Perplexity is starting anew, Google is trying to reinvent itself, and Bing is busy riding the bandwagon.
Now, Ali International has also announced its entry into the game, directly bringing a new product. But unlike everyone else, Alibaba International's AI search cuts into a blank space: the field of in-depth information search that involves more industry know-hows. Alibaba International chose the "global e-commerce" industry that it is best at, with the intention of changing the global procurement process to make it more intuitive and efficient.
At a press conference held in Paris, France on July 31, Alibaba International officially announced the world's first AI-powered B2B procurement search engine. It is understood that this AI procurement search engine will be officially unveiled in September this year.
At the press conference in Paris, Zhang Kuo, vice president of Alibaba International, showed this new product. It proactively understands the natural language of the purchaser and translates it into a professional purchase request; It can also forecast demand and provide recommendations based on global market data for more accurate matching.
The industry believes that this will become the "next Google" in the field of global trade.
"The new AI search engine is not an upgrade on the traditional passive search, it is a new global trade sourcing experience." Zhang Kuo said.
For small and medium-sized business owners engaged in procurement, it can first understand your "big vernacular", automatically translate it into professional procurement terms, and then talk to you step by step, and intelligently integrate the information of the whole network, turn passive search into active understanding, and more accurately understand and even predict your needs.
At the same time, it reconstructs the way information is presented, and can actively help you compare information and recommend the most suitable supplier.
Finally, it can also provide a more complete procurement service, which ultimately helps you complete the entire trade process.
If in the past, Google was only passively responding and matching the information on the web page with the keywords you entered, then this new AI procurement search engine is really understanding the information of the product, understanding the needs of the enterprise, and then actively completing the accurate matching.
The release of Sora once made everyone marvel that "AI is beginning to understand and generate the real physical world", so this AI search engine seems to be beginning to understand the business logic of the real world?
AI makes "interlacing" no longer "like a mountain"
Since the rise of the Internet for decades, "search engines" have been the mainstream way for the public to access all kinds of information.
However, we know that traditional search engines are expensive to obtain information in specialized fields, and searching for expertise on the vast Internet is like "looking for a needle in a haystack".
This is especially true for buyers and sellers of cross-border e-commerce, where the work of discovering business opportunities and determining procurement channels is time-consuming, labor-intensive, and involves a cumbersome data classification process, which is a very complex knowledge system for human beings.
If you don't have some professional knowledge yourself, you don't even know what to search. This may be the traditional meaning of "interlaced like mountains".
People need AI search tools that help them quickly find deeper, more valuable information, and better understand their needs.
The large model technology that is good at instruction following, long context summarization, and content generation has become the key force to leverage this search revolution.
To put it simply, the AI procurement search engine launched by Alibaba International is like injecting expert knowledge in the field of cross-border e-commerce into the general model, providing an "intuitive and natural way of knowledge emergence". Their AI products have learned about 1 billion goods and industries, which is almost the world's largest knowledge base in the B2B trade space.
At the launch event in Paris, Ali International presented the example of Simone, a former professional tennis player. After retiring, she found that the emerging sport of pickleball was very popular and wanted to start a business.
But with a lack of professional knowledge and no contact with global procurement, what should she do first?
The answer is, open this AI search engine of Alibaba International and tell the AI what you think like a chat.
The AI procurement search engine can do in-depth analysis based on all the market insights related to pickleball in the world, and recommend multiple possible entrepreneurial directions, such as pickleball rackets, training shoes, etc. It also comprehensively lists the indicators: fierce competition, off-peak season, market demand, price segment......
After reading the situation researched by the AI, Simone chose to start with pickleball training shoes, and based on her years of experience in playing tennis, she put forward more requirements for the training shoes it designed: to be able to adjust the training mode.
Of course, when Simone came up with the idea, she had no experience in sneaker manufacturing at all, and she didn't know how to adjust the training mode.
It doesn't matter, the AI will actively interpret it: "You actually need to find a supplier that can add weights to the shoes"
Soon, Simone's best supplier of these shoes in the world was revealed:
After that, AI can also understand the information of various products and suppliers it recommends in multiple dimensions, and compare various indicators of different supplies with one click, which makes Simone's choice much easier.
Don't underestimate this step, if you put it in the past, relying on traditional search engines, entrepreneurs or purchasers want to complete this step, you have to click on one web page, find the desired product information in the vast amount of information, and then record it and compare it one by one.
As a result, the benefits of AI Search are evident, and the cost of obtaining valid answers for users can be reduced by orders of magnitude compared to the past.
In addition, this AI procurement search engine will also provide more complete intelligent procurement services in addition to information retrieval.
With the help of Alibaba International's 25 years of deep cultivation and accumulation in the field of digital foreign trade, the AI procurement search engine will also integrate the expertise in all aspects of global trade such as transaction payment and logistics fulfillment, and help complete a full range of services such as communication summary and follow-up delivery like a truly professional human buyer.
Global e-commerce is where AI comes best in
Global e-commerce has always been one of the best application directions of AI technology. The abundant scenarios in the e-commerce industry are an excellent place for AI to be used.
However, the AI procurement search engine launched by Alibaba International is completely different from the simple efficiency improvement tools in the past, and to some extent, it is no longer an AI that simply assists simple work, but is too much like a "super personal assistant".
The realization of this capability is closely related to several technological advancements in the era of generative AI.
As we all know, the most common source of large model pre-training data is the public Internet, and many professional fields have outdated and missing information.
In order to solve this problem, on the basis of the general large model, the industry often uses supervised fine-tuning to update the model knowledge to improve specific capabilities. RAG (Retrieval Enhanced Generation) is another effective method. In simple terms, RAG provides access to a larger knowledge base by retrieving relevant documents and then using them as additional context to perform generation.
It is understood that the AI procurement search engine released by Alibaba International has learned more than 1 billion pieces of commodity and industry information on the whole network during the training stage, especially professional product knowledge. This is also why it can accurately transform the simple "big vernacular" into professional vocabulary in the field of e-commerce, and complete in-depth screening work to help merchants complete some of the shortcomings of professional knowledge.
For Alibaba International, bringing a new cross-border e-commerce experience with generative AI technology will be an important future growth point for its global business. The release of the AI procurement search engine is just one of the links.
As early as November last year, Alibaba International released its first AI product "Aidge", which opened 15 APIs and multiple agent frameworks and supported 18 languages, helping global merchants to overcome language and cultural barriers and improve business results in different countries.
In the past year, Alibaba International has tested AI capabilities in more than 40 scenarios, served more than 500,000 small and medium-sized merchants, and optimized 100 million products. On average, every two months, businesses are doubling their AI calls and are now averaging 50 million calls per day.
Continuous AI investment, abundant AI scenarios, and surging AI demand are the core elements that support Alibaba International's launch of a new AI procurement search engine and transform the global trade experience in a truly AI-native way.
The "Pro Google" of the AI Era
Are you coming?
When generative AI took the world by storm, there was a sobering argument in the industry that all applications deserve to be redone with large models.
As one of the highly anticipated "rework" directions, the value of "AI Search" to improve user experience has been fully proven in practice. From content recommendation to knowledge integration, the threshold for any field is constantly lowering, making it more direct and easy for ordinary people to access information.
After the "redo" for the professional track, the landing value of AI Search is also more concrete. The AI procurement search engine released by Alibaba International is the first to draw the landing form of "B2B Google in the AI era", and sets a sample for the application of generative AI technology in various professional fields.
With the addition of more players in subdivisions, it is conceivable that each industry will have its own "Google" in the future.
The impact of the change in the way information is obtained will be essential. A search-driven revolution may come sooner than we think.