The 2025 China IT Operations Large Model Practice Annual Report, guided by ITSS and compiled by the IT O&M Large Model Standards Working Group, was recently released. CanWay was selected as a leading case for its Agentic Ops practices. As the first industry-level annual report focusing on IT O&M (Operations and Maintenance) large models, it highlights best practices that demonstrate both technical maturity and business value.
01 Industry Recognition
The report provides a comprehensive overview of the development of IT O&M large models in China and selects benchmark cases based on technological innovation, implementation maturity, and business impact. Selected cases have been validated in real-world production environments and demonstrate strong potential for industry-wide adoption across finance, energy, telecommunications, government, transportation, manufacturing, and other sectors.
This recognition demonstrates the maturity of CanWay’s Agentic Ops practices and further validates its leadership in Intelligent IT Operations.
02 From AI Assistance to AI Autonomy: CanWay BlueWhale Agentic Ops Architecture and Practices
CanWay continues to advance Agentic Ops with the goal of evolving AI from an assistant into an autonomous operator capable of executing operations tasks independently. Built on more than a decade of experience in IT operations platforms, the company has developed a four-layer Agentic Ops architecture covering infrastructure, AI capabilities, agent development, and application scenarios.
The foundation is an integrated IT Operations Infrastructure.It integrates Configuration Management Database (CMDB), Full-Stack Observability Center, IT Service Management Center (ITSM), application release and automation capabilities, Disaster Recovery Center, Multi-cloud Management Center (CMP), digital operations, and knowledge management.
These capabilities provide a unified data and execution foundation for AI agents.
The second layer is the Private Ops LLM platform, which supports mainstream foundation models alongside CanWay BlueWhale’s proprietary IT Operations Large Model. It provides centralized model management and delivers efficient, accurate AI decision-making capabilities.
The third layer is the CanWay BlueWhale AI Agent Development Platform, enabling organizations to rapidly build and reuse AI agents through standardized development tools, reducing development costs and accelerating enterprise AI adoption.
The top layer consists of a comprehensive AI agent ecosystem, covering automated inspection, fault diagnosis, IT workflow agents, and standard operation agents, helping enterprises automate daily IT operations and improve operational efficiency.
Key Application Scenarios
The Fault Diagnosis AI Agent establishes a complete automated diagnosis workflow. As CMDB accuracy increased from 60% to 85%, the agent’s false positive rate decreased from over 30% to single digits, enabling more accurate root cause analysis and actionable remediation recommendations.
The ITSM Workflow Agent further evolves AI from an intelligent assistant into an autonomous task executor. Supporting multiple trigger mechanisms—including user conversations, scheduled tasks, and workflow engines—it automates end-to-end IT service processes while continuously improving efficiency.
The Automated Inspection AI Agent performs end-to-end, multi-layer automated inspections. By combining anomaly detection, time-series analysis, and topology analysis, the Automated Inspection AI Agent automatically generates standardized reports and triggers related tickets and alerts, significantly reducing manual inspection efforts.
CanWay introduces the innovative “No Silver Bullet, but a Flywheel” philosophy——highlights that no single technology can solve every complex challenge, but continuous improvement through reinforcing cycles. And aim to leverage AI to accelerate the evolution of IT Operations Governance Agents.
Canway applies this principle to AI-driven operations: high-quality data governance is the foundation—not the outcome—of effective AI. By continuously improving configuration data, monitoring data, and knowledge quality, organizations provide reliable data for AI, while AI further enhances governance efficiency, creating a self-reinforcing cycle of Governance → AI → Better Governance.
As the flywheel accelerates, data quality, AI capabilities, and operational efficiency continue to improve together. In practice, the Configuration Data Quality Agent reduced manual verification workload by over 60%, the Monitoring Operations Agent increased effective alert rates to over 70%, and the Knowledge Base Operations Agent increased fault knowledge retention from below 30% to over 70%.
No Silver Bullet, but a Flywheel
As AI LLM continue to advance, the barriers at the algorithm layer will inevitably decline. Lasting competitive advantage in the AgenticOps era will instead come from assets that cannot be replicated overnight—high-quality IT operations data, a mature MCP tool ecosystem, and a continuously evolving library of AI agents and operational capabilities.
Guided by its philosophy of "The Co-Evolution of IT Operations Governance and AI," CanWay remains committed to helping enterprises build a closed-loop operating model that connects Sense, Decide, Execute, and Optimize. We empower organizations to evolve from AI-assisted operations to autonomous IT operations, unlocking the next leap in intelligent operations.
