How AI is Revolutionizing Observability and Intelligent Operations

During the GOPS Global Operations Conference and R&D Intelligent Technology Summit in Shanghai, a closed‑door industry forum titled “Intelligent Operations for the Future” gathered dozens of experts from finance, technology, energy and other sectors to discuss observability, AIOps and intelligent agents.

1. Challenges and Breakthroughs in Observability

A senior operations expert identified four major challenges: data governance difficulty, system silos, high organizational coordination cost, and vague value measurement. He urged building a business‑centric full‑scope monitoring system with unified data standards and cross‑department collaboration to shift from passive alerts to proactive insights, and called for industry‑wide metric standardization and exploration of large‑model potential in data correlation analysis.

2. Observability Practices for Applications and Intelligent Agents

An intelligent‑operations specialist illustrated, using the financial industry, how cloud‑native and AI‑native architectures exponentially increase operational complexity. He proposed a “golden metric system” covering performance, efficiency and availability, and using digital twin technology for fault simulation and root‑cause location, emphasizing the need for cross‑domain knowledge bases linking business, application and infrastructure.

3. Innovative Practices of Intelligent Operations Platforms

A representative from a financial institution described its intelligent operations platform that integrates metrics, logs, tickets and other data sources to provide risk alerts, real‑time monitoring and fault root‑cause analysis. Leveraging large models, the platform built dozens of operational agents, markedly improving alarm handling and emergency response, and explored AI‑assisted war‑room operations for critical scenarios.

4. Knowledge‑Enhanced AI Operations: From Fault Delimitation to Decision‑Making

An AI‑operations expert explained that AI‑ops has progressed from the DIKW model to a “knowledge‑wisdom” stage, combining expert experience with large‑model reasoning to construct explainable fault‑diagnosis chains, and integrating digital twins to boost decision transparency and trustworthiness in complex operational contexts.

5. The Agent‑Driven DevOps Revolution

A DevOps authority highlighted that large models are driving software development from code writing to natural‑language‑driven processes. Intelligent agents can collaboratively handle requirement analysis, code generation and test verification, promising a human‑machine co‑creation model that greatly enhances development efficiency; balancing workflow orchestration with autonomous agents is key to practical adoption.

Consensus and Controversy

Experts called for industry‑level metric standardization, especially for middleware and databases, and for joint efforts by regulators, leading enterprises and research institutions. Data governance and ROI measurement remain focal points, with suggestions to standardize data interfaces at procurement and validate value through scenario‑based metrics such as fault reduction frequency. Opinions diverged on intelligent‑agent frameworks like LangChain and AutoGen, with some favoring predefined workflows for controllability and others advocating autonomous agents for long‑term potential.

The forum concluded that the core of intelligent operations lies in business‑value orientation and data‑governance foundation, urging continuous cross‑domain knowledge sharing, industry standard formulation and cultural transformation to make AI a true “multiplier” for operational efficiency.