Roundtable Dialogue: Which Track Will the Next Killer AI Product Emerge From? | 2026 AI Partner · Beijing Yizhuang AI+ Industry Conference
The article features a roundtable discussion where industry experts debate the next killer AI product. The consensus is that simple applications based
Deep Analysis
Introduction
The roundtable discussion centers on identifying the next breakthrough AI product that could achieve mass adoption, akin to the smartphone revolution. Participants from companies like Leqi and SenseTime explore whether it will be AI glasses, intelligent agents, or another entry point into the physical world. This conversation reflects broader industry shifts from isolated, chat-based AI to integrated, real-world applications, driven by advancements in multimodal models and hardware innovation.
Key Viewpoints from Participants
Bao Weiji's Perspective on AI Glasses
Bao Weiji, from Leqi, champions AI glasses as the ideal candidate for the next killer product. He argues that glasses are always online, closer to users than phones, and enable longer engagement with the physical world. Key points include:
- Hardware as an entry point: Glasses offer a natural interface for continuous interaction, unlike phones which require active opening.
- Importance of display: A visual interface is crucial for real-time feedback, enhancing human bandwidth for information processing—similar to how reading is faster than listening.
- Ecosystem integration: The long-term value lies not just in hardware sales but in building an operating system (OS) and ecosystem that supports agents and applications, enabling tasks like real-time navigation or price comparison.
Lu Shaoqing's Emphasis on Physical-World Agents
Lu Shaoqing from SenseTime extends the discussion to intelligent agents that operate in physical environments. He highlights:
- Beyond digital confines: Current AI agents are limited to chat interfaces; the next step is agents that can interact with the physical world, understanding context like who is speaking and when to respond.
- Multimodal foundations: SenseTime focuses on multimodal models as the "IQ ceiling" for products, but emphasizes that an AI operating system is needed to manage tools, context, and proactive interactions.
- Real-world collaboration: The goal is agents that function as true assistants—not just in chatboxes but in real-time scenarios, such as providing reminders or handling tasks without overwhelming users.
Debates on Display and Hardware Strategies
The discussion touches on technical and strategic nuances:
- Display vs. no-display glasses: Participants compare glasses with and without displays, noting that displays enable visual feedback for complex tasks (e.g., navigation, inspections), while no-display versions are lighter but rely on audio or external devices.
- Hardware-first vs. ecosystem-first: Bao advocates for a long-term hardware approach with ecosystem expansion, while Lu stresses the need for an integrated system combining models, agents, and hardware entry points like glasses or robots.
Background and Logic Behind the Shift
Limitations of Current AI Products
The conversation underscores a key pain point: most AI applications are "wrapped" around large models (e.g., chatbots), which are short-lived because they lack deep integration into users' lives. The logic is clear:
- Always-online necessity: For AI to be impactful, it must be constantly accessible and connected to physical contexts, rather than requiring manual activation.
- Interaction loops: Real-world applications demand closed loops where AI perce
Disclaimer: The above content is generated by AI and is for reference only.