Deep Integration of Physical Retail Industry with AI Unlocks Vast Value Potential
-
Recently, Alibaba Cloud announced the open-sourcing of its large model Tongyi Qianwen. Around large-model AI, domestic and international internet giants have joined the competition, with AI's influence catalyzing digital transformation across industries.
The revolutionary breakthrough of ChatGPT relies on massive data inputs and substantial support from Microsoft. It is reported that GPT-4 has parameters exceeding the trillion level. In China, large models from Baidu, Tencent, and Huawei are at the hundred-billion parameter level. Processing data of this magnitude undoubtedly requires enormous costs and resources. General large models are the battleground for giants, so where is the opportunity for other companies? In contrast, vertical large models deeply rooted in specific industries, supported by precise industry data, can solve niche problems with just billions of parameters or even fewer high-quality data points, achieving cost-effective and refined solutions. Giants are also collaborating with companies that have accumulated industry data to compete in the industry-specific large model market, offering numerous industry leaders the opportunity to join the vast AI race. Suddenly, large models for education, legal, finance, and other industries have emerged, as these fields are inherently closely related to language and data, making them ideal for large language models.
Additionally, the retail industry, with its massive market share, may be one of the first sectors in the physical economy to deeply integrate with AI. Online, e-commerce has advantages in data accumulation, internet, and AI technologies. Offline, physical retail involves numerous repetitive tasks, vast amounts of data awaiting analysis, mining, and application, and the need to meet diverse user demands. Large models can demonstrate high value in addressing such challenges. Moreover, physical retail, having long been impacted by e-commerce, has advanced further in digitalization compared to many other industries.
This sector has also attracted many giants. For example, 'Dmall,' which specializes in providing digital transformation services for physical retailers, has reportedly partnered with several top-tier AI companies globally to actively develop applications for eight major retail scenarios, including intelligent customer service, industry knowledge bases, and personalized marketing, with weekly iterations. Using Dmall as a case study, we can analyze how companies can seize opportunities to develop vertical industry large models.
Massive data is one of the core competencies. The qualitative leap in AI stems from quantitative accumulation. Although industry-specific large models require far less data than general models, without data sources, they remain 'a clever wife cannot cook without rice.' Data is a key factor determining future development. The ability to obtain massive, high-quality data is one of the core competencies of digital companies.
In recent media reports, Dmall founder Zhang Wenzhong mentioned, 'In the face of the AI revolution, we will further integrate with AIGC, enabling traditional retailers to become active participants in human digitalization and intelligence.' For years, he has advocated for 'comprehensive digitalization' in physical retail, with Dmall providing end-to-end omnichannel retail cloud solutions. This means that vast amounts of data on products, customers, and other aspects across the entire retail process—from consumption to supply—can be stored in Dmall's systems for business analysis, becoming high-value data assets.
By gross merchandise volume, Dmall is currently the largest retail cloud solution provider in Asia. This also means Dmall can integrate big data from domestic and international retail sectors of various formats, leveraging its data advantage.
Incorporating industry best practices to address pain points.
Why does the physical retail industry need AI to address its current pain points? Firstly, consumer demands have become more diverse and personalized, requiring the collection, analysis, and updating of vast amounts of data on users and products. Secondly, to meet these demands, most physical retailers have expanded into omnichannel operations, which complicates supply chain management. Thirdly, the retail industry as a whole is now competing in a saturated market, necessitating more precise cost control.
The necessity of industry-specific large models lies in their ability to address the unique needs and pain points of vertical sectors. Beyond data, these models must incorporate industry knowledge and rules to deliver accurate and effective decisions that drive real-world business outcomes.
From its founders to various departments, Dmall boasts a team of seasoned retail professionals with years of experience. By serving leading retail enterprises such as Metro, 7-Eleven, and Lawson, Dmall has integrated best practices into its systems. Now, by combining these insights with cutting-edge AI technology, Dmall is equipping AI with a comprehensive and advanced knowledge base of the retail industry, making its large models more practical.
Underlying AI development capabilities are indispensable. While general-purpose large models provide a foundation, a company's in-house R&D capabilities remain crucial. Dmall is collaborating with teams from Baidu and Tencent to enhance its R&D efficiency using AIGC technology. Applications are already being tested, ranging from intelligent customer service and personalized marketing to backend retailer management improvements.
Many of Dmall's existing products already leverage AI and big data technologies. For example, automated inventory management uses historical sales data and weather forecasts to predict customer purchasing behavior, optimizing stock levels and reducing overstocking or shortages. Such a robust foundation in model and algorithm development is essential for refining industry-specific large models.
With advancements in 5G and IoT, physical retail will generate even more data, enabling further innovation and upgrades through AI. The future of physical retail will be highly intelligent. Retail is all about details and service, and AI applications hold immense potential, accelerating digital transformation and deepening the integration of the digital and physical economies. The first to develop a dedicated large model for physical retail will gain a significant first-mover advantage by harnessing the Matthew Effect of data applications.