Recently, a leading Middle Eastern operator, together with Huawei, successfully completed the core network prototype validation of the AI Calling assistant “CallingClaw”, built on OpenClaw agent framework. This milestone marks the evolution of AI Calling: from auxiliary functions such as voice recognition and translation to full task execution loops, including flight search, appointment Scheduling and e-commerce queries, bringing new directions for the transformation of operators’ voice services.
Calling as a Service: From “Call Enhancement” to “Call Execution”
Today, most AI Calling applications focus on enhancing the call experience, without fundamentally changing the core value of voice services. CallingClaw signals a new phase: moving from “call enhancement” to “call execution.”
In the prototype validation, users discussed travel plans during a call (meetings, flights, and preparation items). After the call ended, the system automatically completed tasks such as scheduling reminders, flight search and filtering, and product search with cart addition on e-commerce platforms.
This entire process does not interfere with the call itself, forming a seamless loop of “in-call understanding, post-call execution.” The role of the calling interface is rapidly evolving into a service entry point.
Security First: Building a Telecom-Grade “Secure Sandbox” to Protect Privacy and Data Sovereignty
As AI begins to participate in task execution, security and controllability become critical. CallingClaw introduces an innovative “secure sandbox” mechanism, where all call content, user preferences, and generated data are processed with encryption within a trusted network environment.
It strictly adheres to the principles of “data does not leave the network” and “user-authorized access.” User profiling and personalized services operate within a highly secure environment, and users can view and manage the information stored by the assistant at any time, ensuring true data ownership: your data, your control.
A “Growing” Assistant: The More You Use, the Better It Knows
CallingClaw is designed around the core principle: “the more you use it, the better it becomes—and the better it understands you.”
Leveraging the secure, trusted, and reliable nature of the network, combined with OpenClaw’s memory capabilities, the system continuously learns from user-provided information during calls. It safely analyzes communication habits, preferences, and key needs, gradually building a personalized user profile.
Technically, CallingClaw is not a single standalone capability. It integrates network-side semantic understanding, contextual modeling, task structuring and orchestration, and cross-platform service integration—forming a truly evolving, personalized assistant.
The Next Step for Voice Services: From Connectivity to Value Creation
With the deep integration of AI and communication networks and the gradual emergence of the 6G vision, the role of voice services is being redefined. Real-time communication is rapidly evolving toward intelligence.
The validation of CallingClaw on a leading Middle Eastern operator’s core network demonstrates a viable path for AI Calling to move from “assisting communication” to “participating in execution.”
Such innovations can expand the value boundaries of voice services, unlock new monetization paths for network capabilities, and establish a new service entry point for the AI era.
In the 5G-Advanced phase, CallingClaw will further enhance in-call AI orchestration capabilities to support more interactive intelligent services. Looking ahead to 6G, guided by the concept of “AI as a Service,” it will continue to evolve toward ubiquitous collaboration and autonomous decision-making—driving real-time communication from connectivity services to intelligent services.