Summary of Key Points
During the 2026 World Artificial Intelligence Conference, Dr. Xu Hao, Vice President of Qualcomm Global, stated at the Honor sub-forum that "the era of agents has arrived." Agents will completely transform the design logic of consumer devices, shifting from a passive response model where you give commands and they act to a service-oriented model that proactively anticipates your needs. Qualcomm is advancing the adoption of agents across various devices such as smartphones, headphones, cars, and robots by leveraging its low-power sensor hubs, heterogeneous AI computing capabilities, multi-terminal chip architectures, and deep collaborations with partners like Honor. The goal is to create agents that become "intimate assistants" that truly understand your preferences.
Detailed Explanation
1. Agents Are Here: Devices Are Changing From Tools to Assistants
In the past, we used devices as tools commanded by us—e.g., we tapped on the WeChat app to send a message or manually adjusted the volume. In the era of agents, the relationship between humans and devices has evolved into a tripartite interaction involving "human + device + agent." Agents silently observe our behavior and environments behind the scenes and take proactive actions to assist us. For instance, when you enter a meeting room, your phone automatically silences itself; when you emerge from an elevator, it switches to a network with better signal; or when you get in the car, the infotainment system adjusts the seat position and temperature to your preferences. These actions are not prompted by explicit commands but are the result of the agent's proactive predictions. Dr. Xu Hao emphasizes that this will reshape our daily routines, as agents can help us manage our schedules, prioritize important information, and reduce our workload.
2. Hardware Must Keep Up: Two Critical Abilities Are Essential
For agents to provide proactive services, hardware must address two key challenges:
- Low-Power Continuous Sensing: Agents need to collect data (such as heart rate, location, and environment) 24/7 without consuming excessive power. Qualcomm's "sensor hub" serves this purpose, acting like a constantly vigilant "butler" that monitors your status with minimal energy consumption—for example, tracking steps while you walk or recognizing meeting scenarios.
- Powerful AI Computing: Agents require sufficient computing power to function on smartphones. Qualcomm's heterogeneous architecture ensures efficient collaboration between the CPU (for general tasks), GPU (for graphics), and NPU (for AI-specific tasks). The NPU handles AI-related tasks efficiently and efficiently, overcoming computational limitations at the device level.
3. Software Algorithms: The Secret to Making Agents More Understandful
Hardware alone is not enough; software algorithms are crucial for creating intelligent assistants:
- Model Compression and Optimization: Large models can be too resource-intensive for smartphones. Qualcomm has developed techniques to compress them, allowing essential AI tasks (like understanding voice commands and generating responses) to be processed locally, reducing the need for constant cloud connectivity and improving response times.
- Personal Knowledge Graphs: Agents need to learn your habits. For example, if you prefer concise messages, the phone can generate replies in that style; if you often ask a robot to pour you a drink upon returning home, it will remember this preference. These habits are stored in a "personal knowledge graph," serving as the basis for the agent's decision-making—equivalent to providing an "exclusive user manual" for the agent.
4. Agents Are Not Limited to Smartphones: They Can Be Found in Every Device
Qualcomm's chips are compatible with a wide range of devices, from small headphones to large data centers:
- Smart Headphones: They already support simultaneous translation and can filter important voice messages in the future.
- Cars: In-vehicle systems can remember your driving habits and navigation preferences.
- Robots: Home robots can automatically clean and cook based on your routines.
- Data Centers: They enable more extensive AI operations.
5. Ecological Collaboration: Working with Honor to Bring Agents to Reality
Qualcomm and Honor have a long-standing partnership and have jointly launched many smartphone products. In the era of agents, they are continuing to collaborate closely. For instance, Honor's Agentic OS system utilizes Qualcomm's hardware and algorithm technologies. The two companies plan to integrate agent capabilities into more devices, offering users convenient proactive services—e.g., organizing work documents or real-time foreign language translation.
Conclusion
The core of the agent era is proactive service, and Qualcomm's technological strengths (sensors, computing power, algorithms, and multi-terminal support), along with its collaborative ecosystem (including partners like Honor), are driving this transformation. In the future, every device around us could become an intelligent assistant, making our lives and work more convenient and hassle-free.