Baidu (百度) has developed a comprehensive digital human ecosystem centered on its Xiling platform within Baidu AI Cloud. This system enables the creation and deployment of hyper-realistic virtual humans using technologies such as large language models, speech synthesis, facial and voice cloning, and 2D/3D modeling. These digital humans are applied across sectors including finance, education, broadcasting, e-commerce, and customer service. Baidu offers tools for livestreaming, virtual anchors, interactive assistants, SaaS-based creation platforms, and developer kits for content generation. The digital humans are designed to be easily customized, integrated into various scenarios, and used for tasks like virtual hosting, product promotion, and AI-driven interaction.
Robin Li (Li Yanhong, 李彦宏) is Baidu’s co-founder and CEO and has been a prominent internal advocate for treating digital humans as a strategic application layer for Baidu’s AI stack, tying their success less to abstract “one model for everything” ambitions and more to targeted capabilities that convert directly into usable products, especially in commerce and search; he has described an application-driven approach to Baidu’s ERNIE/Wenxin models in which the models are trained to excel at skills that digital humans need in practice—such as instruction following and persuasive, creative script generation—so that digital humans can perform convincingly in livestreaming and e-commerce sales, where dialogue quality and real-time presentation materially affect conversion; in public remarks associated with Baidu’s 2025 developer and product cycle, he has also framed “high-persuasion digital humans” as a multimodal product form built on Baidu’s foundation-model technology and has characterized digital humans not merely as a single feature but as a foundational interaction interface that could generalize across many human–computer scenarios, with Baidu’s Huiboxing (慧播星) presented as a platform vehicle for deploying these capabilities at scale.
Ping Xiaoli (平晓黎) is a Baidu Group vice president and the person reported as leading Baidu’s e-commerce business and its digital-human (数字人) efforts, including responsibility for consolidating internal digital-human teams under the e-commerce line, with reporting that about 15 staff were transferred from the “big search” organization into the e-commerce-aligned digital-human team under her oversight. She has been publicly identified in Chinese media as Baidu Group vice president and “Baidu digital human & e-commerce business head,” and has spoken in that capacity about Baidu’s Huiboxing digital-human livestreaming product (慧播星数字人), describing it as broadly deployed across multiple industries and positioned as a productivity driver for e-commerce-adjacent scenarios. A longer biographical profile published by China State-owned Assets Supervision and Administration–affiliated media describes her as joining Baidu in 2007 and holding product and business management roles across Baidu’s ad network, finance, and mobile ecosystem, becoming Baidu App general manager in 2018 and later moving into e-commerce leadership. She has also been quoted in industry coverage on virtual/digital hosts for commerce livestreaming, describing digital avatars as a way to reduce merchant livestreaming operating costs and enable continuous operation.
Baidu AI Cloud’s “Xiling” (曦灵数字人平台) is a full-featured digital human platform offering 2D and 3D avatar creation, real-time livestreaming, video generation, and dialogue interaction across sectors such as e-commerce, finance, media, education, and government. Powered by large AI models, Xiling supports low-cost, rapid customization and deployment of ultra-realistic virtual humans capable of operating 24/7. It enables businesses to create digital hosts, brand ambassadors, and customer service agents, with integrated tools for content production and marketing automation. Xiling is positioned as a SaaS platform delivering scalable virtual human applications with AI-driven personalization.
(Baidu AI Cloud and Baidu Smart Cloud are identical entities; the former is the official international brand name, while the latter is a literal English translation of the Chinese name Baidu Zhǐnéng Yún (百度智能云). Whether referred to as "AI" or "Smart," the service provides the same integrated suite of cloud infrastructure and advanced AI tools—such as the XiLing digital human platform and ERNIE large language models—positioning Baidu as an AI-first cloud provider for both domestic and global markets.)
https://xiling.cloud.baidu.com
Huiboxing (慧播星) is Baidu's AI-powered digital human live-streaming platform designed for e-commerce. It enables businesses and individuals to create hyper-realistic digital human hosts using just a smartphone and a short video. The system features “one-click live streaming,” “real-person cloning,” and script-based automation to simplify the broadcast process. The platform supports natural expression, voice, emotion, and movement alignment, resulting in digital humans that rival real hosts in persuasiveness and performance. More than 100,000 digital humans have been created through Huiboxing, contributing to a 31% average boost in conversion rates and an 80% reduction in live-streaming costs. It has supported large-scale sales, including during major campaigns like Double Eleven and for rural farmers. Baidu appointed Luo Yonghao as Chief Experience Officer, launched his digital twin, and invested ¥100 million to add 100,000 more digital humans. The platform incorporates Baidu’s Wenxin large model, making it the first full-stack AI digital human live-streaming solution in the industry.
Baidu (百度) Keevx is an AI digital-human video production and localization product positioned for marketing teams, especially cross-border e-commerce operators who need scalable presenter-led content without filming on-camera talent. In a digital-human workflow, Keevx functions as an end-to-end pipeline that can generate avatar-presented short-form marketing videos, reuse or create synthetic presenter assets, and localize output for international distribution by automating dubbing, subtitles, and language adaptation at high scale, with emphasis on rapid turnaround and consistent on-screen delivery. The product framing centers on compressing tasks that typically require multiple specialists—script drafting, voice and subtitle localization, and presenter performance—into a single operator workflow by combining template-driven creation, automated content generation, and digital-human presentation, so the avatar acts as the repeatable front-end “host” for product explanations and promotional messaging across many target markets.