The agentic era of telco AIOps: unifying IT and network operations with human oversight

The telecommunications industry stands at the threshold of a transformative era—one where artificial intelligence operations (AIOps) evolve from reactive, siloed automation to a unified, agentic model. In this new paradigm, autonomous, collaborative AI agents orchestrate both IT and network domains, delivering end-to-end service assurance, self-healing capabilities, and hyper-personalised customer experiences. Yet, as we embrace this future, robust human oversight remains essential for safety, compliance, and trust.

From siloed automation to unified agentic AIOps

Traditional telco operations relied on fragmented tools and manual processes. The agentic era transforms this model: specialised AI agents—monitoring networks, managing IT infrastructure, and handling customer care—collaborate under a master orchestrator. This “master-agent” coordinates cross-domain workflows, enabling agents to plan, invoke tools (APIs, scripts, ticketing systems), and execute actions such as rerouting traffic or initiating repairs. The outcome is near “lights-out” automation for routine tasks, freeing humans to focus on exceptions, strategy, and innovation.

The unified data lake: foundation for intelligent operations

At the heart of this transformation lies a unified data platform—a centralised data lake that ingests real-time telemetry from network probes, OSS/BSS, IT monitoring, CRM, and billing systems. This data lake is enriched by semantic modelling, which links key entities (customer, service, network element, IT asset, slice) and their relationships. By providing a consistent, explainable context, semantic modelling empowers AI agents to correlate symptoms across domains, diagnose root causes, and orchestrate remediation steps that span both IT and network. For sensitive/sovereign data, multiple data lakes with a data mesh approach may be required.

Key capabilities of the unified data layer include:

• Streaming ingestion and normalisation of events, metrics, logs, and topology.
• Semantic joins for precise root cause analysis and impact assessment.
• Feature services for predictive agents (capacity, churn, risk).
• Governed access for compliance, with audit trails and guardrails.

Integrated workflows: IT and network in harmony

Unified AIOps platforms enable seamless collaboration between network and IT agents. Network agents monitor KPIs (latency, packet loss, cell load), optimise configurations, and trigger capacity scaling. IT agents oversee service SLOs, detect anomalies, and remediate issues via integrated tooling. Customer-experience agents leverage the semantic model to link user issues (e.g., streaming jitter) to both network slices and IT microservices, triggering coordinated fixes and proactive customer messaging.

Examples of unified operations:

• A customer’s video streaming issue is resolved by correlating network congestion and IT microservice bottlenecks, triggering both network tuning and IT resource scaling.
• An autonomous agent detects simultaneous errors in a cloud-based billing system and increased packet loss, automatically restarting microservices and rerouting traffic before customers are impacted.
• A DDoS attack prompts the orchestrator to coordinate network rerouting, IT server scaling, and customer notifications for a unified, rapid response.

Event-driven, real-time decision making

Agentic AIOps platforms shift from batch analytics to event-driven operations. Agents react instantly to spikes in dropped calls, security alerts, or traffic surges, executing pre-approved playbooks or proposing options to humans. Predictive agents anticipate capacity bottlenecks or churn risk, while risk-aware agents simulate ‘SLA’ impacts before changes, enabling faster incident resolution and more resilient operations.

Human in the loop: safeguarding autonomy

While automation drives agentic AIOps, human oversight is integral. Telcos set clear policies for actions needing approval—such as major network changes, security-sensitive tasks, high-impact customer decisions, and novel scenarios. Tiered autonomy levels (“in the loop,” “on the loop,” “out of the loop”), approval workflows, and explainability dashboards ensure control where it matters.

Examples:

• AI proposes a network upgrade plan with impact analysis for manager approval.
• Firewall rule changes flagged as high-risk go to compliance with full audit trails.
• Large refunds or contract changes require commercial manager sign-off, with options and impact provided.
• Unusual incidents or conflicting recommendations escalate to human experts for final decisions.

Business impact and outlook

The unified, agentic AIOps model delivers measurable advantages:

• Comprehensive service assurance: End-to-end visibility and control across IT and network domains.
• Proactive, self-healing operations: Minimised downtime and enhanced reliability.
• Personalised customer experience: Anticipatory support and tailored engagement.
• Operational efficiency: Automated routine tasks, enabling teams to focus on strategic priorities.
• Agility and innovation: Faster service deployment and rapid response to evolving business needs.
• Strengthened security and compliance: Integrated governance with human oversight for critical actions.

Conclusion: the path to AI-native telco operations

In the next 3–5 years, leading telcos will evolve into “AI-native” operators, with unified AIOps platforms acting as the central nervous system for both IT and network. By pairing a data lake and semantic model foundation with multi-agent orchestration and human-in-the-loop governance, telcos will deliver adaptive, autonomous services at scale—safely, compliantly, and with exceptional customer experience.