On November 18, 2025, Ant Group launched LingGuang, a novel multimodal AI assistant—the first of its kind in China—capable of generating outputs driven by code execution. This system leverages a modular architecture to decompose user queries into task frameworks, processing information across diverse knowledge domains and modalities in parallel to deliver responses encompassing language, imagery, audio, and data visualizations. LingGuang distinguishes itself with features like “Flash App,” enabling users to create functional applications via natural language prompts in under 30 seconds, and “AGI Camera” for real-time scene understanding; innovations intended to democratize access to Artificial General Intelligence.
LingGuang’s Multimodal Capabilities and Dynamic Content Generation
Ant Group’s LingGuang represents a significant leap in AI assistants, uniquely delivering responses through code-driven, multimodal outputs. Beyond text, LingGuang generates dynamic content including 3D models (like the Pyramids or a Tyrannosaurus rex), animations illustrating complex topics (quantum entanglement), and interactive maps for practical applications. This capability moves beyond simple information retrieval, aiming for intuitive understanding—transforming how users engage with AI and complex data.
Leveraging native AI coding, LingGuang allows even non-technical users to create personalized “flash apps” with simple natural language prompts. These apps, built and runnable within the conversation in under 30 seconds, cover areas like fitness tracking, financial management, and trip planning. This marks a first-of-its-kind general-purpose AI tool empowering broad application development—democratizing access to app creation previously limited to skilled coders.
At its core, LingGuang decomposes user requests into task frameworks, processing them across various knowledge domains in parallel. This modular approach, combined with its coding abilities, ensures logical rigor while generating diverse outputs – from animations to charts – presented in a minimalist style. Ant Group’s CTO He Zhengyu emphasizes LingGuang as a step toward AGI as a “public good,” delivering a personal AI developer to every user.
AI-Driven Application Development and Real-Time Scene Understanding
Ant Group’s LingGuang represents a significant leap in AI-driven application development, moving beyond text-based responses to deliver dynamic, multimodal outputs. This next-generation assistant utilizes a modular approach, decomposing user requests into tasks processed across various knowledge domains in parallel. Crucially, LingGuang can generate functional “flash apps” – personalized tools for tasks like fitness tracking or financial planning – from simple natural language prompts in under 30 seconds, democratizing app creation for non-technical users.
LingGuang’s “AGI Camera” feature showcases advanced real-time scene understanding. By analyzing dynamic footage, the system provides contextual insights and enables on-the-fly image/video editing via voice command. This isn’t simple object recognition; the system interprets complex scenes and responds intelligently, hinting at capabilities beyond current computer vision standards. This level of contextual awareness is powered by Ant Group’s progression toward trillion-parameter-scale AI models.
The innovation hinges on LingGuang’s native AI coding ability, allowing it to translate requests into executable code and generate diverse outputs – from 3D animations to interactive maps. This capability extends beyond simple generative AI; LingGuang aims to be a “personal AI developer,” offering accessible tools for complex problem-solving. Ant Group emphasizes making Artificial General Intelligence a “public good,” positioning LingGuang as a platform to benefit all users, not just experts.
