Google Cloud Adds AI Agents for Telecom Network Data and Voice Operations

Telecom operators seeking more automated network operations are being offered a new route through their data problem. Google Cloud and Future Connections have added two AI agents to Google Cloud’s Autonomous Network Operations framework: one designed to connect fragmented network data and another focused on monitoring voice-service performance.

Before the agent can act

The premise is that telecom data is often split among radio-access, core-network and probe systems, with vendor-specific counters and data formats making it hard for people and AI systems to interpret the whole picture. Google Cloud says that fragmentation can force a separate data-integration project for each new agent.

The new Autonomous Data Steward is intended to be a common data layer for other agents. Powered by Gemini and built with Dataplex Universal Catalog, it uses a zero-copy design: it keeps metadata pointers to large datasets rather than duplicating them.

What the data agent is meant to do

• Turn natural-language requests into the relevant telecom counters and data sources.
• Create and run Python-based extract, transform and load workflows through Dataproc when needed.
• Generate federated queries across domains such as radio performance, core-session failures and probe data.

Up to 70%Potential storage-cost reduction

Google Cloud says the Autonomous Data Steward’s zero-copy architecture can reduce storage costs by up to 70%. This is a company performance claim, not an independently reported result.

A narrower job for voice

The Core Network VoLTE Agent has a more specific remit. It monitors voice-service measures including call-setup success rates, Mean Opinion Scores and CSCF performance, then correlates SIP and Diameter signaling with core-network data to identify potential causes of service problems.

New Zealand operator One NZ has deployed the voice agent for proactive voice-quality management. The announcement does not include measured results from the rollout or describe how much decision-making the agent performs without human involvement.

A framework, not a finished network

Google Cloud says both capabilities are being integrated into its broader Autonomous Network Operations framework. It also says the underlying methodologies are open-sourced so operators can develop and customize their own agent workflows. The move shifts the product pitch beyond an individual diagnostic tool: the practical test is whether operators can make data from different network domains usable enough for several agents to act on it together.