Frost & Sullivan has released the 2026 China General-Purpose AI Agent Enterprise Deployment and Best Practices: Selected Industry Cases of Baidu Dazi. The report defines general-purpose AI agents as intelligent systems capable of multimodal interaction and execution, cross-scenario and cross-domain adaptation, chain-of-thought reasoning and autonomous learning, and goal-oriented delivery, and finds that they are moving from answering questions to delivering results. More than 80% of enterprises already use AI in at least one business scenario, yet fewer than 40% believe AI has had a positive impact on profit. As enterprise demand shifts from individual productivity to organization-wide productivity, the focus of general-purpose AI agents is moving from individual users to large-scale enterprise deployment.
The report identifies five challenges in enterprise deployment: unstable delivery, fragmented systems, a lack of enterprise context, limited understanding of specialized scenarios, and data security and compliance concerns, and finds that deployment advances through four stages: scenario pilots, departmental rollout, process integration and organization-wide scale. Using Baidu Dazi as a case study, the report examines its architecture, capabilities and deployment maturity, and presents eight practice cases across seven industries. Looking ahead, enterprise context infrastructure, deep delivery through forward deployed engineers (FDE), modular skills and professional suites, and trustworthy security governance will be the four trends in enterprise deployment. Enterprises are advised to start with high-frequency shared and data-intensive scenarios, and to define permission granularity and end-to-end auditing before opening up system access.
