Google makes data from the real world more accessible to AI and Training pipelines will love it | Techcrunch

Google is changing its extensive public data in a Goldmine for AI with the debut of the Data Commons Model Context Protocol (MCP) server-value developers, data scientists and AI agents gain access to Real-World statistics using natural language and better AI systems.

Launched in 2018 in 2018, Google’s Data Commons will organize public datas sets of a Sources reachIncluding government investigations, local administrative data and statistics from worldwide authorities such as the United Nations. With the release of the MCP server, this data is now accessible through a natural language, so that developers can integrate it into AI agents or applications.

AI systems are often trained on noisy, non -rewarded web data. Combined with their tendency to “fill in the empty spots” when sources are missing, this leads to hallucinations. As a result, companies AI systems for specific use cases often need access to large, high-quality data sets. By publicly releasing the MCP server for its data community, Google wants to tackle both challenges.

The new MCP server of Data Commons transmits public data sets -from census figures to climate statistics -with AI systems that are increasingly dependent on an accurate, structured context. By making this data accessible through natural language prompts, the release ai aims to ground information in verifiable, real-world.

“The model context protocol lets us use the intelligence of the large language model to choose the right data at the right time, without understanding how we model the data, how our API works,” said Google Data Commons Head Prem Ramaswami in an interview.

MCP was first introduced by Anthropic and is an open industrial stand with which AI systems have access to data from various sources, including business tools, content repositories and app development environments, which offers a common framework for understanding contextual instructions. Since the launch, companies such as OpenAI, Microsoft and Google have adopted the standard for integrating their AI models with different data sources.

While other technology companies investigated how they could apply the standard to their AI models, RamasWami and his Google team started investigating how the framework could be used to make the Data Commons platform more accessible earlier this year.

Google also works together with one campaign, a non -profit organization that focused on improving economic opportunities and public health in Africa, to launch one data agent. This AI tool uses the MCP server to surface tens of millions of financial and health data points in normal language.

One campaign approached the Google’s Data Commons team with a prototype implementation of MCP on its own adapted server. That interaction, Ramaswami told Techcrunch, was the turning point that the team led to build a special MCP server in May.

However, experience is not limited to one campaign. The open nature of the Data Commons MCP server makes it compatible with every LLM, and Google has offered different ways for developers to get started. A sample agent is available through the Agent Development Kit (ADK) in a Colab NotebookAnd the server is also directly accessible via the Gemini Cli or any MCP-compatible client using the PYPI PACKAGE. Sample code is also given on a Github -Repository.