live wire
▸Snowflake agent evals can pin judge versions; unpinned metrics may roll forward and shift scoresSnowflake▸ThoughtSpot makes Spotter 3 default for new instances; it requires download rights and always shares query data with an LLMThoughtSpot documentation▸MotherDuck adds image classification and object location to prompt_jev(), returning typed SQL outputs with confidence scoresMotherDuck▸BigQuery conversational analytics adds Montréal and Toronto; mixed-region agents still default to globalGoogle Cloud Documentation▸Fabric enables Azure OpenAI Copilot access by default, but OpenAI-subprocessor and cross-region processing/storage switches remain offMicrosoft Learn▸Snowflake Cortex Agents now continue by default when named tools are inaccessible, emitting warnings instead of failing the runSnowflake documentation▸Databricks Genie One can use group memberships to tailor answers, while Unity Catalog remains the access-control boundaryDatabricks documentation▸Fabric data agents can query Power BI models with Read alone; Build and workspace access are unnecessaryMicrosoft Learn▸Fabric data-agent Git configs expose column flags, but selecting a table still includes every columnMicrosoft Learn▸Google Data Agent Kit offers remote Spark kernels, but notebook exports can still move results into local Pandas or ArrowGoogle Cloud Documentation▸Cube says production analytics agents need explicit partial-result states, caller-filtered model search and durable artifactsCube▸Thinking Machines names Bridgewater behind ReViSQL-K2.6, citing 92.97% SC-16 accuracy at 15% of frontier-model costThinking Machines Lab▸AWS says a mismatched data/tool/prompt release cut its Field Sales agent pass rate from 78% to 38% in pre-productionAWS Artificial Intelligence Blog▸NL2SQL Multi-Agent ships evaluation artifacts across BIRD, CSpider, Archer and KDD DataSpace—but no software licenseGitHub▸Snowflake agent evals can pin judge versions; unpinned metrics may roll forward and shift scoresSnowflake▸ThoughtSpot makes Spotter 3 default for new instances; it requires download rights and always shares query data with an LLMThoughtSpot documentation▸MotherDuck adds image classification and object location to prompt_jev(), returning typed SQL outputs with confidence scoresMotherDuck▸BigQuery conversational analytics adds Montréal and Toronto; mixed-region agents still default to globalGoogle Cloud Documentation▸Fabric enables Azure OpenAI Copilot access by default, but OpenAI-subprocessor and cross-region processing/storage switches remain offMicrosoft Learn▸Snowflake Cortex Agents now continue by default when named tools are inaccessible, emitting warnings instead of failing the runSnowflake documentation▸Databricks Genie One can use group memberships to tailor answers, while Unity Catalog remains the access-control boundaryDatabricks documentation▸Fabric data agents can query Power BI models with Read alone; Build and workspace access are unnecessaryMicrosoft Learn▸Fabric data-agent Git configs expose column flags, but selecting a table still includes every columnMicrosoft Learn▸Google Data Agent Kit offers remote Spark kernels, but notebook exports can still move results into local Pandas or ArrowGoogle Cloud Documentation▸Cube says production analytics agents need explicit partial-result states, caller-filtered model search and durable artifactsCube▸Thinking Machines names Bridgewater behind ReViSQL-K2.6, citing 92.97% SC-16 accuracy at 15% of frontier-model costThinking Machines Lab▸AWS says a mismatched data/tool/prompt release cut its Field Sales agent pass rate from 78% to 38% in pre-productionAWS Artificial Intelligence Blog▸NL2SQL Multi-Agent ships evaluation artifacts across BIRD, CSpider, Archer and KDD DataSpace—but no software licenseGitHub
nl2sql.ai
newsProduction data agents

The UN turns official statistics into a natural-language and MCP data service

The UN System Data Commons joins official datasets behind one searchable graph—and gives AI agents a route to the same governed source material.

Before-and-after view of siloed data turning into a unified governed graph.
AI-generated illustration
By The News Desk· Sep 17, 2026the quick take — two AI hosts go live when you do

The United Nations system has launched a new version of data.un.org built for both human questions and machine agents. The UN System Data Commons presents official statistics through natural-language search, interactive exploration and an open-source Data Commons foundation. Google’s launch post says the platform replaces months of manual reconciliation across separately formatted datasets with one interconnected, AI-ready knowledge graph.

The interface is conversational; the product is a shared data layer

A user can ask questions such as how rural clean-water access relates to school attendance and receive relevant data plus visualizations. The live portal also supports structured browsing by geography and themes including health and education. Google describes the service as a unified environment that aligns metrics, timelines and geographic boundaries across previously siloed datasets.

That distinction matters for practitioners. This is not a chatbot laid over one warehouse table. The difficult work is upstream: aligning metrics, timelines and geographic boundaries from multiple institutions so that a natural-language layer has consistent objects to retrieve. Google says UN statisticians and technical experts validate every dataset made available through the system.

MCP extends the same source to research agents

The launch also exposes the data for agentic workflows through open standards including the Model Context Protocol. According to Google, agents can fetch authoritative figures from the Commons, connect information across domains, and assemble charts, infographics or draft reports. Google explicitly cautions users to inspect underlying sources before citing critical figures, even when the retrieved data is grounded and verified.

For data-agent teams, that combination is more instructive than the search box alone. The platform pairs a public conversational interface with a reusable machine-access layer rather than forcing every assistant to scrape pages or rebuild statistical mappings. It also keeps provenance visible: the goal is to accelerate research assembly, not erase the review step.

The graph is still expanding

Google says the UN system plans to add more datasets over the coming year and aims to include 80% of UN system statistical datasets by 2027. The launch was supported by Google.org through the UN Foundation, while the underlying Data Commons technology is open source.

The immediate lesson for enterprise analytics teams is architectural: reliable natural-language access begins with a shared, governed representation of the data. MCP can distribute that context to multiple assistants, but it does not replace semantic alignment, provenance or human verification.

sources

Filed by The News Desk. Corrections: desk@nl2sql.ai · Our standards →

comments · 0

    Comments are moderated before they appear. Your email is used once to confirm it is you — never shown, never sold. Corrections and questions get an answer from the desk when we have one.

    more from the desk