Dominic Jainy stands at the forefront of the next great shift in telecommunications, bringing a wealth of experience in artificial intelligence, machine learning, and the burgeoning world of blockchain. As an IT professional who has spent years dissecting how complex systems communicate, he offers a unique perspective on the industry’s transition from rigid, manual oversight to the fluid, self-correcting nature of Level 5 autonomous networks. With the recent advancements in configuration management and agentic AI, Jainy provides a roadmap for how global operators can move beyond simple automation to create truly intelligent, self-healing infrastructures.
The following discussion explores the critical evolution of Operational Support Systems into what are now being called Autonomous Intelligence Systems. We delve into the persistent challenges of configuration drift and operational silos, examining how the integration of orchestration, assurance, and analytics creates a closed-loop architecture. The conversation also highlights the shift from massive, centralized data lakes toward a more agile, federated data approach that leverages digital twins and graph views to provide the context AI agents desperately need to make trusted decisions.
Manual interventions and fragmented tooling often lead to configuration drift and increased operational risk. How do these traditional “human-in-the-loop” errors specifically compromise the stability of modern, multi-vendor networks?
The real danger in today’s complex environments isn’t usually found in the initial setup or “day-zero” deployment, but rather in the long-term lifecycle of the equipment. We often see a scenario where a superuser logs directly into a router to make a quick fix, an action that immediately bypasses central tracking and creates a discrepancy between the intended and actual state of the network. This manual intervention is a primary driver of configuration drift, which acts like a silent infection that spreads across fragmented tools and operational silos. When you have multiple vendors involved, these small, unrecorded changes manifest as a flurry of alarms and faults that leave engineers scratching their heads, not knowing exactly where to look for the root cause. By the time a service degradation is noticed, the original change might be weeks old, making the audit trail nearly impossible to reconstruct without a unified management layer.
Blue Planet has recently introduced a new Configuration and Change Management platform. In what ways does this technology act as a foundational layer for achieving the industry’s “holy grail” of Level 5 autonomy?
To reach Level 5 autonomy, we have to move beyond treating automation as a series of isolated tasks and instead embrace a complete closed-loop architecture. The CCM platform is significant because it stops the “infection” of configuration errors from spreading into the wider operational environment by providing a stable foundation of network intelligence. It’s not just about pushing a script to a device; it’s about connecting orchestration, assurance, analytics, and automation so they work in concert. This foundation allows the system to understand the “why” behind a configuration change and to see how that change impacts service levels three levels up the chain. Without this ability to track and audit attributes over time across multiple endpoints, any attempt at higher-level autonomy would be built on a house of cards.
There has been a long-standing debate about the “single data model” versus distributed data. Why is the current shift toward leaving data where it lives, rather than pulling it into a centralized lake, proving to be more effective for AI integration?
For decades, the industry chased the dream of a single, unified data model, but the reality is that every vendor wants to differentiate, and that dream has largely remained out of reach. Moving petabytes of live telecommunications data into a centralized repository is not only prohibitively expensive but often results in data that is obsolete by the time it arrives. Instead, we are seeing a shift toward a “graph view” or a digital twin approach, where we pull data from distributed infrastructure into connected cloud platforms only as it is required. This allows us to maintain a live, evolving view of the network where we can plan and test changes without the massive overhead of a centralized database. AI agents thrive in this environment because they can access the most current, relevant information from inventory systems and service records without being bogged down by stale, moved data.
How does the introduction of “agentic AI” change the firepower available to telcos compared to the more traditional, task-based automation we’ve seen in the past?
Agentic AI frameworks represent a massive leap forward because they can process both structured and unstructured data—like technical documents, engineering records, and service tickets—at a lightning pace that humans simply can’t match. Traditional automation follows a “if this, then that” logic, but these new frontier models can actually sift through mountains of information to compare configurations and build proactive playbooks. They act as an intelligent layer that can plan, optimize, and manage networks that are becoming increasingly complex due to the sheer velocity of new devices and sites. This “firepower” is becoming urgent because the scale of modern networks has grown beyond the capacity of human-centric management, making these agents essential for understanding the root causes of performance issues before they escalate.
You’ve mentioned that AI agents are only as good as the context they are given. Can you describe the practical steps involved in building a “decision framework” that allows these agents to learn from previous outcomes?
Building a decision framework is about much more than just feeding an AI a stream of telemetry; it’s about mapping the operational reasoning behind every diagnosis and resolution. We want to create a system where future agents can look back at a previous outage, see what specific actions were taken, and understand whether those interventions actually achieved the desired outcome. This involves tracing changes as symptoms of outages and ensuring that if a parameter is reset due to performance degradation, that information is fed back up the chain to the orchestration engine. By preserving these decision paths, we provide the AI with the context it needs to make “trusted” decisions, rather than just guessing based on raw data. This context is what transforms an efficiency tool into a reliable operational layer that can truly govern the network.
As we move away from traditional Operational Support Systems, what does the future look like for “Autonomous Intelligence Systems” in the daily life of a network operator?
The transition from OSS to Autonomous Intelligence Systems means that the software is no longer just a “support” tool but has become the brain of the operation. In the daily life of an operator, this means moving away from firefighting individual alarms and toward managing a cohesive knowledge base where planning, inventory, and compliance are all working in harmony. The system becomes a self-optimizing entity that can detect a symptom, trace it back to a configuration change made weeks ago, and remediate it without needing a manual ticket. We are essentially moving into an era where the “connectivity discipline” is about the seamless flow of data between silos, turning the entire network into a responsive, intelligent organism.
What is your forecast for the adoption of Level 5 autonomy in the telecommunications sector?
I believe we are going to see a phased but accelerating adoption of Level 5 autonomy, starting with specific, high-risk areas like configuration management before expanding to the entire network loop. Within the next few years, the pressure of managing 5G and beyond—with its massive increase in sites and services—will make human-led management a physical impossibility, forcing operators to trust agentic AI layers. We will likely see “pockets of autonomy” where the closed-loop systems for assurance and remediation become the standard, eventually merging into a single, autonomous intelligence framework. The winners in this space will be those who stop trying to build bigger databases and instead focus on building the best context and governance for their AI agents to act upon.