Dominic Jainy stands at the forefront of the autonomous network revolution, bringing a wealth of experience in artificial intelligence and machine learning to the complex world of telecommunications. As an expert who has spent years dissecting the intersection of blockchain and automated systems, he offers a unique perspective on how global carriers are moving away from rigid, manual operations toward fluid, agentic AI ecosystems. His insights are particularly timely as the industry shifts from localized optimizations to overarching, self-healing infrastructures that promise to redefine the user experience.
The following discussion explores the strategic evolution of network automation, specifically focusing on the transition from domain-specific tools like RAN Guardian to comprehensive platforms such as MINDR. We delve into the mechanics of multi-agent systems, the move toward Level 4 autonomy, and the ambitious productivity targets that are currently shaping the future of connectivity.
The shift from manual network planning to an agentic approach has enabled the analysis of 100,000 events compared to just 1,000 previously. How does this massive leap in scale fundamentally change the way a modern network breathes and responds to its environment?
The transition is nothing short of a paradigm shift because it moves us from a reactive posture to a predictive one. In the old days of manual planning, we were essentially trying to look through a keyhole; our engineers could only focus on about 1,000 major events, which meant we were constantly “behind the wave” when traffic spiked unexpectedly. By deploying an agentic approach that can handle 100,000 events, the network gains a form of peripheral vision that was previously impossible. It allows the infrastructure to sense the subtle vibrations of a city, identifying potential congestion before it even manifests as a dropped call or a buffering video. This scale is vital for preventing roughly 24 million minutes of poor customer experiences, ensuring that the network doesn’t just survive high-demand scenarios but thrives within them.
When we look at the live deployment of systems like RAN Guardian, there is a fascinating synergy between three distinct agents. Could you walk us through how these agents collaborate during a major public event to ensure the signal stays strong for everyone?
The beauty of the RAN Guardian system lies in its modular choreography, where three specific agents work in a tight feedback loop. It begins with the Scanning agent, which acts as the eyes of the operation, pulling data from social media and public news to detect where crowds are gathering in real-time. Once an event is identified, the Monitoring agent takes over to assess the technical weight of that crowd, calculating the impact on capacity and signal interference. Finally, the Remediation agent steps in to physically and logically reshape the network, executing actions like antenna-tilt adjustments or carrier aggregation to funnel capacity exactly where it’s needed. Since going live in Germany back in late 2025, this system has autonomously triggered over 100 remediation actions, proving that it can manage the frantic energy of a stadium or a festival much faster than any human operator could.
As the industry moves toward MINDR, the scope is expanding from the radio access network to a much broader automation platform. What does this cross-domain intelligence look like in practice, and why is it necessary for reaching Level 4 autonomy?
MINDR represents the evolution from “specialized silos” to a “unified nervous system” for the entire network. While RAN Guardian was a hero in its own domain, MINDR is designed to correlate signals across multiple nodes and domains, providing a holistic view that is essential for Level 4 autonomy. It’s no longer just about fixing a single tower; it’s about specialized agents across the network reasoning together to find the root cause of an issue that might span from the core to the edge. This modular observability allows us to move away from isolated use cases toward a scalable autonomy where the system anticipates issues rather than just reacting to them. By the end of 2026, as we roll out the initial MVP in Germany, we will see a platform that doesn’t just follow a script but actually understands the service-level implications of every incident it encounters.
The platform follows a rigorous four-step process: signal correlation, root-cause analytics, remediation, and continuous learning. How does the “human-in-the-loop” philosophy fit into this highly automated workflow, especially when dealing with critical infrastructure changes?
Even in an increasingly autonomous world, trust must be built on a foundation of safety and human oversight. In the third step of the MINDR process—remediation—we categorize issues based on their impact; if an incident is low-risk, the agent is empowered to fix it autonomously and verify the result within predefined guardrails. However, for any change deemed “critical,” we maintain a strict human-in-the-loop requirement, ensuring an expert approves the action before the machine executes it. This creates a powerful synergy where the AI handles the exhausting “noise” of minor tickets, while humans are reserved for high-stakes decision-making. The fourth step, continuous learning, ensures that every outcome—whether human-approved or fully autonomous—is fed back into the system to refine future responses, creating a virtuous cycle of improvement.
With targets like 90% automated ticket handling and incident resolution in under an hour, the productivity gains are significant. Beyond the numbers, how do these efficiencies translate into a tangible difference for the person holding a smartphone?
While the 30% productivity improvement and the 15-minute issue qualification targets are vital for our internal operations, the real victory is the “invisible” network. For the end user, these efficiencies mean that problems are often identified and resolved before they even notice a dip in service quality. When we can resolve incidents in less than an hour—compared to the hours or even days it might have taken manually—we are preserving the digital lifeblood of our customers. Whether it’s a surgeon using a remote connection or a student taking an online exam, the speed of agentic AI ensures that the connection is resilient and reliable. Ultimately, autonomy isn’t just an engineering goal; it’s a commitment to a seamless customer experience where the technology works so well that it becomes transparent.
What is your forecast for the future of autonomous networks?
I believe that by the time we move beyond the scaling phase of MINDR in the next couple of years, the very concept of a “network outage” will begin to fade from the public consciousness. We are heading toward a future where networks are biological in their adaptability, possessing the ability to self-heal and reconfigure their architecture in milliseconds to bypass failures. As we integrate more specialized agents that can reason across complex, multi-vendor environments, the role of the network engineer will shift from being a “troubleshooter” to being a “curator” of AI strategy. My forecast is that the agentic layer will soon become the standard for all global infrastructure, turning the internet into a truly proactive utility that anticipates human needs before we even express them.