Baidu EVP Shen Dou: Industrial agents to drive enterprise 'Task Autonomy' phase
Editorial responsibility
- No named human review is recorded for this page.
- Source reporting is collected, normalized, translated or condensed automatically when needed.
- Automatically published source-backed update
On September 20, Shen Dou, Executive Vice President of Baidu Group and President of Baidu Intelligent Cloud Business Group, stated that industrial agents will propel enterprise operating systems into a 'Task Autonomy' phase (Industry 3.0), moving beyond information digitization and process softwareization. He argued that agents' value lies in creating new products, services, and growth opportunities, not just cost reduction. Shen outlined that roughly 20% of daily tasks can be handled by general-purpose agents, 30% by professional agents, and 50% by custom agents built on enterprise-specific rules, data, and processes. Baidu's general-purpose agent, Baidu Dazi, can coordinate these agents and existing systems. To support this, Baidu Intelligent Cloud launched the Baidu Dazi Development Platform and Ecosystem Open Platform, opening underlying capabilities. Shen emphasized that agent deployment requires robust AI cloud infrastructure for security and stability, noting that Baidu's full-stack AI serves over 80% of central state-owned enterprises and 100% of systemically important banks. He characterized the implementation as an organizational transformation requiring synchronized adjustments at resource, tool, and organizational levels, creating a mutually reinforcing cycle between frontline demands and chip/cloud/model progress.
Source report
September 20 — Shen Dou, Executive Vice President of Baidu Group and President of Baidu Intelligent Cloud Business Group, stated that the value of agents lies not only in reducing costs and improving efficiency, but more importantly in helping enterprises create new products, services, and growth opportunities. Industrial agents, he said, will drive enterprise operating systems from information digitization and process softwareization toward a stage of "task autonomy."
Evolution of Enterprise Operating Systems
Shen noted that enterprise operating systems have already passed through two stages—information digitization and process softwareization—yet tasks are still primarily performed by humans. Industrial agents will propel enterprises into the "Task Autonomy" phase of Industry 3.0, enabling systems to autonomously complete certain tasks and codifying the working methods of outstanding employees into capabilities that organizations can reuse repeatedly.
This transformation is not about starting from scratch. Rather, the data, tools, processes, and experience accumulated by enterprises will serve as the foundation for agents to understand business operations and complete tasks. Implementation can begin with tasks that have clear value and well-defined boundaries, gradually expanding over time.
Task Distribution and Agent Types
Regarding specific enterprise tasks, Shen introduced the following breakdown:
- ~20% of daily operational tasks can be handled by general-purpose agents
- ~30% of specialized tasks can be completed by professional agents
- ~50% of core tasks require custom agents supported by the enterprise's own business rules, data, and processes
As a general-purpose agent, Baidu Dazi can coordinate calls to professional agents, custom agents, and existing business systems to form closed-loop results. To support this, Baidu Intelligent Cloud has launched the Baidu Dazi Development Platform and the Baidu Dazi Ecosystem Open Platform, opening up underlying capabilities to help enterprises build their own industrial agents.
Security and Infrastructure
Shen emphasized that once agents enter enterprises, security and stability become even more critical, requiring robust AI cloud infrastructure support. "The competition among agents is essentially a competition of AI clouds," he said.
At the Agent Infra layer, Baidu Intelligent Cloud provides over 150 models. Through Agent Cloud, it offers a stable, secure, and traceable runtime environment for complex tasks. Currently, this full-stack AI capability serves:
- More than 80% of central state-owned enterprises
- 100% of systemically important banks
Organizational Transformation
"The implementation of industrial agents is essentially an organizational transformation driven by technological penetration, vertically penetrating enterprise operating systems, IT systems, and underlying resources," Shen said. "Enterprises need to make synchronized adjustments at the resource, tool, and organizational levels."
As agents delve deeper into real-world business scenarios, frontline demands will reverse-drive continuous progress in chips, clouds, and models, forming a mutually reinforcing industrial cycle that opens new growth spaces for enterprises.
Source: Shanghai Securities News · China Securities Network
Source
东方财富网-产业经济Eastern
Part of this Story
Baidu executive says industrial AI agents will drive enterprise task autonomy era