AI Integration in Live E-commerce: The Development Direction of Digital Human Live Streaming
-
With the advancement of generative AI, digital humans are increasingly appearing in short videos, customer service, and now venturing into live-streaming e-commerce. On August 8, Qianxun, founded by Wei Ya and her husband, launched its AI digital human business and a one-stop AI intelligent live-streaming service platform, further expanding the digital human live-streaming sector.
In fact, besides Qianxun, many platforms and MCN agencies in the industry are attempting to integrate AI into live-streaming e-commerce scenarios. Given the rising costs of traffic and human resources in live-streaming e-commerce, digital human live streaming has become a direction explored by many players.
In China, digital human hosts are primarily used in store live-streaming. AI digital humans possess professional skills, emotional stability, and tireless performance, enabling businesses to accumulate replicable, manageable, and iterable digital assets. This technology will also bring about a 'technological democratization' revolution, making high-quality store live-streaming services accessible beyond major brands, while helping businesses reduce costs.
According to tests by Baidu Intelligent Cloud's Xiling platform, a newly launched digital human live-streaming room featuring an AI digital host selling smartwatches achieved sales of 13,000 yuan in its first 33-hour broadcast, with total costs under 100 yuan.
Industry experts agree that duration is a key metric for live-streaming e-commerce productivity. Currently, digital human hosts in China often collaborate with real hosts in shifts.
AI Digital Humans Make Live E-commerce More Accessible
The popularity of AI large models has reignited interest in virtual digital humans. At the recent 20th China Digital Entertainment Expo (ChinaJoy), virtual digital humans and AIGC (generative AI) were hot topics, with companies showcasing new possibilities across industries.
On July 30, Huawei Cloud announced the launch of its Pangu Digital Human Model, enabling users to drive digital humans via text, voice, or video for creative tasks.
In 2020, virtual digital humans like Vanke employee Cui Xiaopan, Douyin beauty influencer Liu Yexi, and the hyper-realistic KOL Ling sparked discussions, but interest waned as the metaverse cooled.
Practicality is the keyword in this resurgence. Companies have learned from past 'hype without conversion' experiences, focusing more on commercial applications. In March, Wanxing Technology released 'Wanxing Broadcast,' an AIGC tool for marketing videos. In June, 360 launched its '360 AI Digital Human Plaza' with over 200 virtual roles. In July, Xinyi Technology debuted a commercial digital human platform based on video large models.
Large Models Lower Production Barriers
On a computer screen, a female host fluently introduces cereal in Russian. Interestingly, on other screens, she simultaneously speaks Spanish, English, German, Dutch, and French, with identical gestures and tone. These videos are used for cross-border live-streaming on TikTok.
'AI digital human live-streaming has attracted waves of clients,' said a representative from a Fujian cross-border e-commerce incubator. Hiring foreign-language hosts is challenging, so clients seek digital hosts that boost sales affordably. Top-tier digital hosts now cost under 5,000 yuan per month for 24/7 broadcasting.
'Affordable and effective' sounds simple but was hard to achieve. Unity China CEO Zhang Junbo noted that many digital human creators were essentially graphics rendering firms, with high costs for frame-by-frame animation or model cleanup. Mass-produced digital humans often lacked uniqueness, failing to meet personalized needs.
Large models present the possibility of significantly reducing digital asset production costs. A product manager from SenseTime's digital human division stated that their AI-powered digital human video generation platform can automate the entire workflow from video upload, data processing, model training, model conversion to deployment. This reduces manual processing time in traditional video production by 90% and model training time by 60%, enabling completion of various customized digital human training within just one day. Large models can also help virtual digital humans quickly establish conversational relationships across different scenarios and create industry-specific digital humans.
While live-streaming e-commerce is seen as a booming industry, its internal processes previously relied heavily on manual operations with low efficiency. To help merchants improve efficiency, Qianxun launched its digital strategy in 2020, creating the industry's first all-in-one AI live-streaming SaaS platform - Link System, providing comprehensive solutions for SMEs. In recent years, technologies like AI and metaverse have been increasingly applied in live-streaming e-commerce, with virtual hosts gradually gaining public acceptance. Qianxun quickly realized it could provide digital services to brands and SMEs through AI digital human technology.
The era of AI digital humans upgrading brand live-streaming has arrived. AI digital humans offer professional competence, emotional stability, and tireless performance, helping enterprises accumulate replicable, manageable, and iterable digital assets. As this technology develops, high-quality store live-streaming will no longer be exclusive to major brands. Technological innovation can help businesses reduce costs while giving physical commerce digital wings.
Regarding opportunities AI digital humans bring to merchants: traditional brand live-streaming rooms require high operational costs (150,000-250,000 yuan monthly for personnel, equipment, and venue). With AI digital humans, brands can establish their own live-streaming rooms for just thousands of yuan monthly. Moreover, metrics like viewing duration, conversion rates, and transaction volumes often outperform human hosts.
The deep integration with live-streaming e-commerce has become an important application scenario for AI. As an industry leader, Qianxun has proactively built strong digital capabilities and explored innovative AI applications to help merchants seize new opportunities and empower SMEs to break through.
During its rapid growth phase, Tmall was Alibaba's absolute focus. However, with changing circumstances and Taotian Group's new strategy emphasizing 'user-first, ecosystem prosperity, and technology-driven' approach, Alibaba needs to shift focus back to Taobao to better serve SMEs. Focusing on industrial belts means prioritizing SMEs and their cost-effective products, requiring Taobao Live to implement the 'price competitiveness' strategy from Taotian's 2023 five major strategies.
Previously, Taobao had two live-streaming formats: store broadcasts from Tmall flagship stores and influencer-led broadcasts like Li Jiaqi's. Neither format suited industrial belt merchants with sales needs, most of whom weren't on Tmall. Another reality is that many SMEs cannot afford to train their own hosts. Thus Taobao created a new system providing space for 'amateur hosts' to promote industrial belt products.
For example, Taobao's industrial belt live-streaming entered Yunnan province this year, focusing on four major industrial belts (flowers, fruits, tea, and health products), each with 100 billion yuan output value. The platform also covers more scattered agricultural products. While many hosts migrated to Douyin and Kuaishou for better traffic in recent years, increasing competition is driving hosts to reconsider and migrate to more suitable platforms.