Tech Mahindra (NSE: TECHM) has introduced a new Multi-Modal Network Operations Large Language Model specifically tailored for telecom networks. This model has been developed in collaboration with NVIDIA AI Enterprise and AWS Cloud infrastructure, and is designed to enhance network observability, automate anomaly detection, and enable autonomous network operations across telecom environments.
This advanced model leverages the Llama 3.1 8b instruct model, customized extensively for telecom, by training it on large-scale network datasets. By combining generative AI and agentic AI frameworks, the model can handle structured data such as events, alarms, and counters, along with unstructured data like logs, MOPs, SOPs, images, text, and marketing materials. This allows for proactive issue resolution and improved service quality, supporting autonomous networks at L4 and above maturity levels.
Advancing Towards Autonomous Networks in Telecom
The telecom industry has increasingly deployed AI use cases, but these efforts have largely been transactional. To achieve true operational efficiency, telecom networks need to embed AI capabilities into their core processes. Through this collaboration, Tech Mahindra, NVIDIA, and AWS aim to help telecom operators unlock operational excellence by harnessing AI-driven capabilities across network operations.
The Multi-Modal Network Operations Large Language Model is part of Tech Mahindra’s network automation platform, netOps.ai. It also incorporates the Tech Mahindra Optimized Framework, known as TENO, and integrates NVIDIA AI Enterprise components such as NVIDIA NeMo™ and NIM microservices. This advanced AI model works in tandem with AWS services including Amazon Elastic Container Registry (Amazon ECR), Amazon Elastic Compute Cloud (Amazon EC2), and Amazon Elastic Kubernetes Service (Amazon EKS). These elements together form the backbone of Self-Driving Networks, enabling Zero x and Self x capabilities.
Key Benefits of the Large Telco Model for Network Transformation
Manish Mangal, Chief Technology Officer, Telecom & Global Business Head, Network Services at Tech Mahindra, highlighted the importance of transitioning to autonomous networks. Through this partnership with NVIDIA and AWS, Tech Mahindra is leading the development of the Multi-Modal Network Operations Large Model, aimed at improving security, automating network management, and enhancing operational efficiency. By embedding AI into daily operations, telcos can significantly lower operational costs while creating more agile and resilient networks.
Initial Use Cases: Intelligent Observability and Proactive Anomaly Resolution
The first development phase of the Multi-Modal Network Operations Large Language Model focuses on driving operational efficiency through Intelligent Observability. Two primary AI-driven use cases will be introduced:
Dynamic Network Insights Studio: This AI-powered observability platform offers a 360-degree unified view into network performance, delivering deep insights for AI teams, network operations, and C-suite executives.
Proactive Network Anomaly Resolution Hub: This auto-resolution system uses advanced AI techniques to detect and resolve network anomalies — including alarms and events — autonomously, with no need for manual intervention.
Enabling Fully Autonomous Networks with Large Telco Models
Chris Penrose, Vice President of Telco Business Development at NVIDIA, emphasized the significance of large telco models that understand the network language. According to Penrose, the launch of the Multi-Modal Network Operations Large Language Model based on NVIDIA AI Enterprise offers a transformational opportunity for telecom operators to accelerate their shift towards fully autonomous networks. These models serve as foundational frameworks for deploying multiple AI agents capable of delivering AI-accelerated operations across the entire network.
Core Components of the Solution Architecture
The solution architecture driving the Multi-Modal Network Operations Large Model consists of three key components:
- Efficient Data Ingestion: Seamless collection of network data, including events, alarms, logs, and structured datasets.
- Data Curation and Model Customization: Refining AI training through customized datasets tailored for telecom-specific scenarios.
- Automated Action Implementation: Accelerated execution of corrective actions to ensure swift resolution and service restoration.
Driving AI Innovation and Expanding Future Use Cases
With global AI investments in telecom rapidly growing into multi-billion-dollar levels, this collaboration underscores Tech Mahindra’s commitment to AI innovation in telecom network operations. By combining intelligent automation, deep learning, and multi-modal AI models, Tech Mahindra aims to redefine how telecom operators manage network performance, ensure service reliability, and achieve operational excellence.
Looking ahead, Tech Mahindra plans to extend the capabilities of the Multi-Modal Network Operations Large Language Model beyond network operations, applying it to additional business use cases across the telecom ecosystem.
