JD.com Unveils Ambitious AI System, Oxygen, for E-Commerce
JD.com Retail has officially introduced its innovative AI architecture system for e-commerce, known as Oxygen, marking a significant milestone in the industry. Notable enhancements are expected in the upcoming October update of the JD.com app, featuring smart shopping solutions like the “Aigou” guide.
Despite these advancements, the AI e-commerce landscape largely remains in its nascent stage, characterized by ongoing experimentation and exploration.
According to a report from the STAR Market Daily News published on September 28 by reporter Huang Xinqi, the integration of artificial intelligence into e-commerce is progressing rapidly, with the anticipation that by the second half of 2025, this integration will be more pronounced.
As e-commerce giants such as JD.com and Taobao seek to harness the AI wave, the profound amalgamation of large-scale models and their specific business operations has emerged as a primary focus.
The debut of Oxygen serves as a critical advancement for JD.com Retail. The forthcoming major upgrade to the JD.com app will not only unveil the shopping guide “Aigou,” but will also incorporate immersive virtual reality 3D product displays.
Hu Xi, Senior Vice President of JD.com Group and Chairman of the JD.com Group Product Committee, articulated that the firm’s JoyAI large model has led to the development of several specialized predictive, generative recommendation, and multimodal models (VLM).
These sophisticated models transcend mere technical applications; they embody the “brain” facilitating intelligent decision-making processes within e-commerce.
To cultivate a pioneering intelligent e-commerce platform, JD.com plans to establish an array of functional intelligent agents.
These will range from product display agents to procurement, sales, customer service, and marketing agents, all collaborating synergistically to propel the e-commerce sector’s intelligent transformation.
“We have rolled out the multi-agent collaboration framework, OxyGent, alongside the OxyGent-9N-xLLM large model inference engine. More groundbreaking technologies are set to be unveiled in the near future,” Hu Xi noted.
Further investigations by the STAR Market Daily News indicate that JD.com is intensifying its focus on product presentation techniques, particularly within XR and 3D domains.
The company is currently exploring methods to leverage 3D technology for showcasing beauty products, enabling virtual trials, and executing full virtual product broadcasts during live streams, allowing consumers a more immersive shopping experience.
Hu Xi acknowledged the primary obstacle pertaining to these technologies lies in scalability and commercial viability. The intricate processes involved in creating and scripting a 3D model can be both time-consuming and exorbitant, posing sustainability challenges.
To tackle these hurdles, JD.com has initiated the development of an AI-driven 3D editing platform founded on extensive models, aimed at expediting 3D content creation and enhancing scenario integration.
This initiative allows users to generate requisite 3D content through natural language interactions, thus facilitating large-scale production of 3D product presentations.
In addition to JD.com, other platforms like Taobao and Meituan are hastening their AI e-commerce initiatives. Taobao, for instance, has introduced features such as “AI Assistant,” AI Universal Search, and AI Price Comparison, optimizing user shopping efficiency via iterative dialogues for product discovery, bespoke recommendations, and price evaluations.
Similarly, Meituan has launched a dedicated consumer-facing intelligent agent, “Xiao Mei,” aimed at enhancing food delivery services and local recommendations, grounded in users’ historical preferences.
Zhang Yi, founder and CEO of iiMedia Research, remarked on the differentiated strategies employed by e-commerce platforms regarding AI integration. Each company’s approach aligns with its individual strategic trajectory and unique corporate ethos, driving forward competitive differentiation.
JD.com, with its OxyGent framework, champions a comprehensive intelligent model. It addresses traditional search inefficiencies on the consumer front while asserting influence over industry standards and establishing an open-source intelligent framework on the supply chain side.
Conversely, Alibaba prioritizes holistic consumer decision-making support by capitalizing on its ecosystem’s synergy, including platforms like AutoNavi and Ele.me. Meanwhile, Meituan’s focus on local services allows it to leverage AI transformation effectively, with its Xiao Mei AI accentuating merchant recruitment and fulfillment capabilities as a competitive advantage, he elaborated.
From Zhang Yi’s perspective, the AI-driven e-commerce domain is still emerging. Current participants predominantly engage in exploratory activities.
The future will center on enhancing multimodal interactions, synthesizing cross-scenario data, and establishing new industry standards through AI transparency and user privacy measures to steer the technological direction of key players, he elucidated.
When addressing the challenges of implementing large-scale models in the e-commerce context, Hu Xi shared insights with reporters from The Science and Technology Board Daily.
He highlighted model hallucination as the predominant challenge, complicating efforts to achieve deterministic outcomes.
“Presently, intelligent agents are more adept in scenarios laden with uncertainty, such as customer service and sales forecasting. In contexts demanding precise results, reliance on intelligent agents may not be advisable,” he stated.
Hu Xi candidly remarked that JD.com adopts a practical perspective while advancing the deployment of intelligent agents internally.
“The potential inherent in intelligent agents is noteworthy; however, feasibility must govern their implementation. It would be premature to assert that this paradigm will dominate future applications—at least in the immediate term.”
Source link: News.futunn.com.
