At Baidu World 2026, Robin Li introduced the concept of an “intelligent organization”: companies break work into tasks, assemble temporary teams of people and AI agents to complete them, and turn experience into reusable organizational assets. We believe the direction deserves more attention than the slogan, because a task-based model directly changes job roles, performance evaluation, and the boundaries of employees’ work.
What this is
“Intelligent organization” is Baidu’s term for AI’s deeper integration into how enterprises collaborate. Unlike simply adding an AI assistant to existing software, the model envisions breaking roles into tasks and forming teams of people and AI agents—software agents capable of autonomously invoking tools to achieve defined goals. Once a task is completed, the experience is written into the organization’s knowledge base and continuously refined through performance evaluation and review mechanisms.
Li compares AI to a new source of electrical power: replacing the machines is not enough; the factory must be rebuilt. For enterprises, the core of this model is not buying more models, but redesigning workflows, permissions, performance evaluation, and the ownership of experience.
Industry view
The design has obvious appeal: newcomers can draw on historical experience, managers can allocate resources according to outcomes, and agents can take on repetitive work. But the concerns raised in the article are sharper. Once job boundaries disappear, employees may shift from “completing their duties on schedule” to “waiting for tasks at any time.” Once experience is distilled into generic cards, the original contributor may be left with only a vague record of their contribution.
We believe the biggest risk is not that AI will replace a particular role, but that the organization will appropriate the results of employees’ labor without clear rights and protections. If tasks keep increasing and performance records mainly serve to prove that workers are interchangeable, the intelligent organization could become a more efficient system of labor pressure. Whether tasks are voluntary, whether employees have offline time, how experience is credited, and who can use performance data should therefore be settled before deployment.
Impact on regular people
For enterprise IT: system transformation will shift from deploying models to connecting tasks, knowledge bases, permissions, and performance systems, while addressing logging, data security, and human review mechanisms.
For individual careers: employees will rely more heavily on reusable experience and the ability to collaborate through tasks. At the same time, job descriptions may weaken, and whether personal expertise becomes a visible asset will depend on company rules rather than years of service alone.
For consumer markets: in the short term, users will first notice faster service responses and fewer repetitive steps. The real impact will depend on whether companies convert efficiency gains into better prices and services or pass them on as higher hidden costs.