On July 17, 2026, the World Artificial Intelligence Conference (WAIC 2026) officially opened in Shanghai. Under the theme “Intelligent Partners, Co-creating the Future,” the conference brought together top global AI researchers and industry leaders. At the opening day’s main forum, Yin Qi, chairman of StepFun and chairman of Qianli Technology, delivered a keynote speech titled “When AI Agents Enter the Physical World,” systematically laying out his deep reflections on the AI industry’s current development stage, technological trends, and societal impact.

Reflecting on his 15-year AI entrepreneurial journey, Yin expressed amazement at the industry’s transformation. “AI entrepreneurship has evolved from a niche track into a major global consensus,” he said. “Today’s AI researchers and entrepreneurs stand on the shoulders of generations of scientists’ continuous exploration, while also benefiting from the accumulated progress in information technology, mobile internet, cloud computing, semiconductors, and new energy. Being able to participate in this epochal transformation is an extraordinary privilege.”

Model Capabilities Cross Critical Threshold, Programming Becomes New Benchmark

Addressing the current pace of AI technological evolution, Yin offered a clear assessment: model capabilities are crossing a critical threshold in 2026. AI has advanced from being able to execute tasks lasting only a few seconds to working independently for tens of hours, and the industry now stands at the foothills of the Artificial General Intelligence (AGI) summit.

In measuring AI capability leaps, Yin placed particular emphasis on programming. He noted that, following natural language, programming is becoming a crucial benchmark for measuring AI capability advancement. This assessment aligns with the industry’s heightened focus on code generation and autonomous programming capabilities, signaling that AI is accelerating from “understanding the world” toward “transforming the world.”

As agent models increasingly integrate with next-generation terminal products, intelligence is moving from digital screens into the real physical world. Yin believes AI agents will evolve from chatbots into the smallest productive units capable of perception, decision-making, and task execution. Future work scenarios will undergo fundamental changes: engineers, designers, and researchers will each have dedicated AI agents, enabling a single individual to possess the capabilities of an entire team.

Three Structural Transformations: New Systems, New Carriers, New Networks

In the core section of his speech, Yin proposed that the AI agent wave will drive three major structural transformations, corresponding to the restructuring of operating systems, terminal form factors, and network architectures.

At the new systems level, agent capabilities are jointly determined by model capabilities and agent operating system capabilities. Yin introduced the key concept of “Agentic OS,” noting that this operating system—by connecting models with data, tools, interfaces, and devices—will determine how far agents can go. This implies that future technological competition will depend not only on model parameter scale or reasoning levels, but also on who can build a more efficient and open agent operating environment.

At the new carriers level, terminal design philosophy will shift from “human-centered” toward “human-machine symbiosis.” Yin painted a picture of cross-terminal intelligence: computers, phones, cars, and robots will become different “bodies” for the same agent across different scenarios, supporting continuous cross-device presence, free migration, and collaborative task execution. This “one brain, multiple bodies” architecture will fundamentally break the fragmented user experience currently existing between devices.

At the new networks level, Yin proposed the forward-looking concept of an “A2A network” (Agent-to-Agent network). This network will connect humans and agents, with agents possessing their own identities, capabilities, and credit, autonomously seeking partners, organizing collaboration, and completing transactions. This network architecture is positioned as critical infrastructure for the agent economy, hinting at a future parallel collaboration network composed entirely of AI agents.

Order Restructuring: From Capability Leaps to Governance Exploration

While envisioning technological prospects, Yin did not shy away from the profound challenges that agents entering the real world will bring. He emphasized that agents entering the physical world bring not just capability leaps, but also order restructuring. The industry must collectively answer a series of critical questions: On whose behalf do agents act? Who bears responsibility for the consequences of their actions? How can identity trustworthiness, permission controllability, and behavioral traceability be ensured?

These questions strike at the core challenges of AI governance. Yin called on the industry to explore new technological systems, governance mechanisms, and social contracts to address challenges such as ambiguous accountability and accumulated security risks that may arise following large-scale agent deployment.

In closing, Yin expressed his core belief about human-machine relationships: “The future is not a world where machines replace humans, but a world where humans and intelligence co-evolve. When AI agents enter the physical world, every individual’s potential will be amplified tenfold.”

Background and Industry Context

Yin appeared at WAIC in a dual capacity—as chairman of both StepFun and Qianli Technology. StepFun, a significant player in China’s large model arena, has long focused on AGI technology research and development; Qianli Technology specializes in intelligent terminals and robotics. This cross-domain combination precisely印证了 his speech’s core proposition of “accelerating integration between models and terminals.”

From an industry perspective, 2026 marks a critical turning point as the global AI industry shifts from the “hundred-model war” toward “application deployment.” Multiple leading companies have already released agent-related products and strategies, with tech giants including Google, Microsoft, and Meta making heavy investments in agent technology. The three-layer framework of “new systems, new carriers, new networks” proposed by Yin provides a systematic perspective for understanding the evolutionary path of the agent industry.

Market analysts note that the extension of agents from digital space into the physical world will directly drive innovation across industry chain segments including terminal hardware, operating system upgrades, and new network protocol development. In particular, concepts such as “Agentic OS” and “A2A networks” may provide new investment narrative directions for capital markets.

Notably, Yin’s emphasis on governance issues also reflects a growing industry consensus. During this year’s WAIC, multiple sub-forums centered on AI safety, trustworthiness, and governance, with regulatory representatives and industry figures engaging in in-depth dialogue on accountability boundaries in the agent era.

As of press time, neither StepFun nor Qianli Technology has disclosed specific product launch timelines related to the speech content. However, Yin’s address at the WAIC opening forum is already viewed by the industry as a significant strategic declaration from both companies in the AI agent arena.