The One Campaign, a global non-profit that lobbies to end extreme poverty and preventable diseases, has launched an agentic AI data-query platform with Google, allowing academics, researchers, policymakers and others to compare global health financing data for decision making.

Named the One Data Agent, the platform has a particular focus on Africa.

According to Prem Ramaswami, head of the Data Commons project at Google, the platform was built using the Model Context Protocol, which allows agentic AI applications – a user-facing chatbot in this instance – to connect to the Google Commons database of global health financing data to facilitate user queries.

“The One Data Agent is a user-friendly platform for exploring and downloading health financing data. Who should use it? Journalists, policymakers, NGOs, governments and many others,” Ramaswami said in a media briefing on Thursday.

“If our (Google’s) mission is to make this data universally accessible and useful, we need organisations like the One Campaign that can actually take this and turn it into interesting stories that can then push policy, aid and other systems so that we can have an impact on the ground.”

Initiated in 2018, Google’s Data Commons project aims to create a single, reliable source for the world’s public data. Ramaswami said the project has achieved three key objectives since its inception.

Common knowledge graph

These objectives include a common knowledge graph, which has solved the standardisation problem caused by different countries and organisations reporting on the same concepts in different ways. This previously made comparisons across markets difficult, requiring tedious human intervention to draw these parallels before an analysis could be performed, often with the help of a spreadsheet.

A second innovation is the creation of a natural language interface backed by the work of large language models to “take the infinite space of human speech and be able to connect that to this finite space of indicators and variables”.

Read: Wits professor named in Time’s list of 100 most influential people in AI

This specialised LLM for querying a single, trusted data source has many advantages over generic models that scour the internet in that it has a complete understanding of the terms used in the domain being queried and does not introduce any external data sources. Ramaswami said this “takes the hallucination problem” common to large language models “down to zero”.

The third innovation achieved by Data Commons is allowing third-party data sources to overlay their own data over Google’s database via a Google Could instance, allowing applications like the One Data Agent to use both Google and the One Campaign’s historical data to fulfil its objectives. Similar projects are under way between Google Data Commons and the United Nations to make more public data easily available and queryable at a global scale.

Google updates the data at least once a year and sources it from public agencies like Statistics South Africa. Ramaswami said Google’s Data Commons is developing partnerships with such bodies to try and get more frequent updates.

Some example queries given in a demonstration of the One Data Agent include how much is spent on medical goods, in shillings, per person in Kenya; and how much each Nigerian household spends on out-of-pocket medical expenses.

In each instance, the One Data Agent was able to produce accurate results in seconds. Comparisons with generic tools like ChatGPT working over the web would often yield incorrect results.

“As someone who spent 20 years writing briefings and trying to explain this stuff, this is a game changer. The speed at which you can get this analysis, you can focus your time on actually convincing people that this stuff matters and you can reach lots more people,” said David McNair, executive director of global policy and strategy at the One Campaign. — © 2025 NewsCentral Media

Get breaking news from TechCentral on WhatsApp. Sign up here.