Recently, the 30th GOPS Global Operations & Maintenance Summit concluded successfully in Shanghai. Leveraging its pioneering five-layer agent architecture, mature full-stack O&M capabilities proven in production, and abundant real-world industry case studies, the Lerwee OPS Agent has been honored as the 2026 AI O&M · Agent Application Benchmark Product.

The GOPS Global Operations & Maintenance Summit is one of the most influential industry events for R&D, operations and digital intelligence in China. This year’s summit focused on the transformation of O&M paradigms driven by Agentic Ops, bringing together top domestic O&M specialists, enterprise technical leaders and mainstream technology vendors to explore practical paths for intelligent O&M to move from concept to production in the large model era. The benchmark product selection adopted multi-dimensional evaluation criteria covering technical architecture innovation, product maturity, industry implementation outcomes and domestic IT adaptation capability. This award represents high recognition from the industry for Lerwee’s technical strength and market value in the O&M agent track.

As a pioneer of O&M agents, Lerwee has 16 years of deep expertise in O&M technology. It launched the open-source community initiative for the O&M sector back in 2010, building a community with more than 300,000 followers and growing into one of the world’s most active O&M technology communities. Its services span a full spectrum of industries including finance, transportation, aerospace, healthcare, education, telecommunications, manufacturing, energy and government agencies. The company has accumulated over 1,000 successful cases for government and enterprise clients, with its business footprint extending to multiple countries and regions at home and abroad.

Enterprise IT environments are growing increasingly complex today. With multi-cluster cross-region deployment, massive heterogeneous hardware and software, and coexisting cloud-on-premises hybrid architectures, enterprises commonly face pain points such as fragmented O&M data, difficult root cause analysis for faults, overwhelming alert storms, disconnection between business and IT operations, and heavy reliance on senior expert experience. Traditional AIOps solutions mostly stay at the alert analysis stage and struggle to form a complete closed loop spanning perception, assessment and autonomous remediation.

The Lerwee OPS Agent features an original five-layer O&M agent architecture: Perception, Planning, Memory, Brain and Action. It delivers a progressive five-step capability chain: Intelligent Discovery – Full-Stack Observability – Business Deconstruction – Intelligent Analysis – Decision & Execution, bringing the brand-new Agentic Ops O&M paradigm into real production. It shifts O&M from passive alert firefighting to a new autonomous O&M model featuring proactive perception, independent assessment, controlled execution and continuous iteration.

• Perception Layer: Built on the foundation of Lerwee Monitoring, it supports dozens of collection protocols. Its proprietary adaptive technology enables automatic device adaptation, covering servers, network devices, storage, virtualization, containers, databases, middleware, IoT environmental monitoring equipment and multi-cloud environments. Compatible with over 500 vendors and more than 8,000 device models, it realizes automatic discovery and full-stack observability of all-domain IT resources. Combined with business insight capabilities, it automatically generates business topologies, supports end-to-end business probing and SLO service quality monitoring, so teams can clearly understand not only infrastructure status but also the actual operational health of services.
• Planning Layer: Powered by Lerwee CMDB, it implements all-domain asset modeling, CI lifecycle maintenance and automatic relationship synchronization. It supports 3D data center views and business application architecture views, and unifies management for assets, changes, spare parts and compliance audits. It provides the agent with complete context of the IT environment, enabling AI to comprehend the complex dependencies between enterprise IT assets and business services.
• Memory Layer: It establishes a multimodal memory system that integrates knowledge graphs, historical fault cases, time-series monitoring metrics and document vector data. It codifies decades of O&M expert experience accumulated within enterprises, separating short-term session memory from a long-term knowledge base to equip the agent with continuous learning capabilities. Every fault remediation can be solidified into reusable O&M knowledge assets.
• Brain Layer: Serving as the decision-making hub of the agent, it integrates LLM large language models, RAG retrieval-augmented generation, CoT Chain-of-Thought reasoning and a Skill Marketplace. It goes beyond simple Q&A: it can decompose complex O&M problems and invoke the O&M skill library for multi-step reasoning and analysis. The platform comes with abundant built-in CoT O&M scenarios, covering high-frequency use cases including database fault diagnosis, host capacity risk assessment, middleware hidden hazard inspection and zombie resource identification. Custom scenarios are also supported to continuously expand the intelligent O&M capabilities.
• Action Layer: It connects to network management platforms, automated job engines and flexible ITSM workflows. It delivers a full set of capabilities including configuration backup, traffic analysis, batch job execution, workflow orchestration and ticket management. Under security governance and graded command auditing, it enables controlled remediation after fault assessment, closing the complete loop of Discovery – Analysis – Remediation – Ticket Completion, making the agent capable of understanding, reasoning and taking action.

The Lerwee OPS Agent is not a standalone AI module decoupled from the underlying O&M foundation. It is deeply rooted in Lerwee’s O&M product portfolio refined over more than a decade. Large models are natively and deeply integrated with monitoring, CMDB, automation and ITSM. It avoids impractical AI gimmicks and prioritizes usability in real production environments, domestic IT adaptation, and manageability, controllability and auditability. It supports full-stack adaptation for domestic hardware and software as well as complex heterogeneous hybrid cloud environments, balancing security, reliability and intelligence.

Winning this benchmark product award recognizes Lerwee’s long-term strategy of independent R&D and focus on real-world O&M scenarios. It also validates market recognition of the new-generation Agentic Ops paradigm.

Moving forward, Lerwee will keep iterating the capabilities of the Lerwee OPS Agent, enrich the library of CoT O&M scenarios, deepen industry implementation, leverage the power of its open-source community ecosystem, and work together with government and enterprise clients to accelerate the transformation from traditional O&M to intelligent autonomous operations. The goal is to bring O&M agents into production at scale and unlock the productivity value of IT operations.