Google announced two separate initiatives this week expanding how AI agents can access and act on real-world data, launching Model Context Protocol (MCP) support for both a new UN statistical database and its Google Home smart device ecosystem.
The United Nations, in partnership with Google, unveiled the UN System Data Commons, built on Google’s open-source Data Commons platform, allowing users to query UN statistics through natural language and enabling AI systems to connect directly via MCP. João Pedro Azevedo, UNICEF’s chief statistician, cited an unpublished UNICEF study finding that AI models including OpenAI’s GPT-4o, Anthropic’s Claude, and Google’s Gemini averaged only 21.2% accuracy when answering questions about global development indicators, often failing to provide usable answers or returning inconsistent results across repeated queries. Shantanu Mukherjee of the UN Statistics Division said the platform represents a significant advance in connecting data across UN agencies and making it AI-ready. Google.org contributed $2 million toward the platform’s infrastructure, with Prem Ramaswami, who leads Google’s Data Commons team, noting the system traces statistics back to original sources and cautioning that human review remains necessary before publishing AI-generated conclusions.
In other Google news, the company rolled out early access to an MCP server for Google Home, allowing AI agents such as Claude, ChatGPT and Google Antigravity to control connected devices, review camera summaries, and build custom dashboards using natural language. The feature supports Nest devices and Matter-compatible products, and requires users to configure a Google Cloud project before connecting their preferred AI agent. Access begins rolling out to Google Home Premium Advanced subscribers in the US, with Google soliciting feedback from early adopters before considering broader availability.
