As telecom operators evolve from connectivity providers into digital service platforms, AIOps is rapidly becoming central to how networks are managed, optimized, and scaled. The industry conversation has moved beyond dashboards and analytics toward a far more ambitious objective: building networks that can analyze, decide, and act in real time with minimal human intervention.
But realizing that vision requires far more than deploying AI models. The real transformation lies in redesigning operational workflows, breaking down organizational silos, and building trust in autonomous systems.
At the upcoming Network X Americas 2026, I will be joining the TIP panel discussion, AI for Telecom Networks in Practice, alongside industry leaders including Kristian Toivo (Telecom Infra Project), Kaniz Mahdi (AWS), Rob Soni (AT&T), Bernard Bureau (Telus), and Prakash Sankara (Reliance Jio). The discussion will explore how large language models and agentic AI are enabling intent-driven optimization and orchestration across telecom networks.
The challenge of adopting AI in telecom is often framed in terms of technology, but technology is only part of the story. Open communities and shared ecosystems are making AI models, frameworks and tools more and more available. What remains difficult is changing the way organizations work around them.
For most operators, organizational is nearly 70 percent of the AIOps journey. Networks are inherently cross-domain environments with different tools, priorities and processes for RAN, core, transport, OSS and cloud teams. Without that alignment, AI could just be another advisory tool layered on top of existing workflows.
True AIOps happens when operators completely rethink operational models – moving away from human-led escalation processes to systems that can analyze, execute and optimize autonomously within defined guardrails.
The transition to autonomy doesn't happen by changing the whole network at once. Successful operators focus on high-impact use cases that tie directly into specific business pain points.
Energy efficiency is a case in point. Operators then identify all data pipelines associated with the use case, map which domains are involved, and establish decision points, execution layers and operational guardrails. From there, cross-functional teams work together to ensure AI-driven actions are coordinated network-wide.
This is critical because, in the best of circumstances, the most valuable AIOps deployments are not limited to a single domain. A RAN optimization decision can impact the behavior of the core network, the utilization of the transport, and the customer experience simultaneously. Without a repeatable framework for collaboration, each AI initiative risks being a disconnected experiment rather than a contributor to long-term operational maturity.
Many organizations started their AI journey with co-pilot models that improved engineering productivity and operational efficiency. These systems still have value because they keep humans in the loop while helping teams validate AI outputs and building confidence in automated recommendations.
However, co-pilots are fundamentally different from autonomous AIOps. Humans still make the final call-in co-pilot environments. AI systems in autonomous environments are transitioning to closed-loop operations where they can sense conditions, take actions, and learn from results with minimal human intervention.
This distinction becomes critical when outages occur. Outages may be unavoidable, but unmanaged operational chaos is not. The co-pilot sees a problem and can suggest how to fix it. To respond dynamically in real time with reduced recovery time, limited impact, and continuous optimization of future actions, true AIOps works within set policies and guardrails.
Trust is the key in the transition from human assisted AI to AI making autonomous decisions. That trust is built over time and is dependent on the criticality of the use case. Operators typically start with low-risk scenarios with limited impact on customers, allowing AI systems time to prove their reliability. With increasing confidence, autonomy is pushed to more complex areas of operation. But transparency is the precondition for trust. As telecom networks evolve into more autonomous agent-to-agent systems, operators risk creating opaque “black box” environments where AI systems interact without enough visibility.
That is why explainability, governance, and orchestration are becoming foundational prerequisites for AIOps maturity. Operators require a single pane of glass to monitor models, data pipelines, and decisions, ensuring all automated actions align with the broader network intent. Defining the operational context is just as important: where AI can act, under what conditions, and when to hand back control to humans. Like autopilot systems in aviation, autonomous networks still require human oversight for high risk or exceptional scenarios.
The operators that are going to win with AIOps are not going to be the ones that deploy the most AI models. They will be the ones building the reusable, transparent, and scalable operational frameworks that enable autonomy to scale safely across the network.
Meet us at Network X Americas 2026 (May 18-20) in Dallas to explore how Rakuten Symphony solutions across OSS, Cloud, Open RAN, and ecosystems are enabling carrier-grade reliability, AI-driven optimization, and true autonomous operations while helping operators reduce energy consumption and OPEX to unlock new growth opportunities.
Connect with me on LinkedIn and look forward to seeing you in Dallas!
