Agentic AI-RAN Demoed at MWC

- SoftBank's Large Telecom Model (LTM) is a specialized generative AI trained on vast amounts of the company's own network data, design, and operational expertise. In tests, it has shown over 90% accuracy in predicting optimal configurations for cellular base stations, reducing the time for these adjustments from days to minutes. - The demonstration is part of a broader industry initiative formalized through the AI-RAN Alliance, launched at MWC in 2024. Founding members include SoftBank, Northeastern University, NVIDIA, Microsoft, and Samsung, who aim to advance radio access network (RAN) technology through the integration of AI. - This type of "intent-based" network translates high-level goals into autonomous actions. For example, an operator can state a business objective, and the agentic system will automatically adjust network slices and resources to meet the required service levels without manual configuration. - Competitors are also advancing agentic AI for networks; Nokia and AWS, for instance, demonstrated an agentic AI-powered network slicing solution at MWC 2026. Their system allows the network to autonomously respond to real-world situations, such as traffic surges during major public events. - Northeastern University brings deep academic research to the partnership, including its work on Colosseum, the world's largest radio frequency network emulator, and its leadership in the Open6G+AI initiative, both funded by the National Science Foundation. - The "agentic" component in AI-RAN signifies a shift from simple automation to autonomous systems that can reason, plan multi-step actions, and adapt to unforeseen events. This enables the network to proactively manage itself, such as predicting and preventing congestion or equipment failures. - Enterprise use cases for LangChain, mentioned as a key player in the agentic AI space, often involve creating workflows that connect LLMs to company-specific data sources and APIs. This allows agents to perform tasks like summarizing internal documents, answering questions against a private knowledge base, or automating customer support resolutions. - For telecom operators, the business driver behind AI-RAN is to improve capital efficiency and unlock new revenue. By making the RAN an intelligent and programmable edge computing platform, they can offer new AI-driven services with the ultra-low latency required for applications like autonomous vehicles, industrial robotics, and AR/VR.