PIB TrackerMCP server links AI tools to official statistics
The statistics ministry has connected AI applications to verified government data through a Model Context Protocol server riding on the e-Sankhyiki portal.
What happened
- The Ministry of Statistics and Programme Implementation (MoSPI) told the Rajya Sabha that the Government has introduced a Model Context Protocol (MCP) server to link Artificial Intelligence tools directly with official statistical databases.
- The server lets AI applications and platforms reach verified Government data held on the e-Sankhyiki National Statistical Portal, instead of relying on scraped, second-hand or possibly outdated figures.
- The earlier Beta version exposed 7 key statistical products — the Periodic Labour Force Survey (PLFS), the Consumer Price Index (CPI), the Annual Survey of Industries (ASI), the Index of Industrial Production (IIP) and the National Accounts Statistics among them.
- That coverage has now been extended to 21 products available on the e-Sankhyiki portal, widening the slice of official data an AI tool can query through the channel.
- The stated aim is to make India's official statistics machine-readable and AI-accessible, so that conversational and analytical AI systems can answer questions using authoritative numbers at the source.
- The disclosure was made by the Minister of State (Independent Charge) for Statistics and Programme Implementation, Rao Inderjit Singh, in a written reply in the Rajya Sabha on 23 March 2026.
Background & context
A Model Context Protocol (MCP) is an open standard for connecting AI models — especially large language models and the assistants built on them — to external tools and data sources through a uniform interface. Rather than every application inventing its own bespoke connector, an MCP "server" publishes a defined set of capabilities (data resources and callable functions) that any MCP-aware AI "client" can discover and use. In plain terms, it is a standard plug between an AI tool on one side and a data system on the other. MoSPI's deployment is notable as one of the first uses of this protocol by an Indian government body to expose official data, which is why it carries exam weight as a named technology rather than as a routine IT notice.
The data on the far side of that plug lives on e-Sankhyiki ("sankhyiki" is the Hindi/Sanskrit-rooted word for statistics), MoSPI's National Statistical Portal. e-Sankhyiki is the ministry's single-window platform for disseminating official statistical datasets and time-series in a structured, downloadable form, built around a "data lake" architecture and Application Programming Interface (API) access so that the same numbers can serve a researcher, a dashboard or, now, an AI client. The MCP server is best understood as a new front door onto that existing portal: e-Sankhyiki holds and serves the statistics; the MCP server translates those holdings into the protocol an AI assistant can natively consume.
MoSPI itself is the nodal ministry for the country's statistical system. It was formed in 1999 by merging the Department of Statistics and the Department of Programme Implementation, and it houses the National Statistical Office (NSO) — the result of a 2019 merger of the Central Statistics Office (CSO) and the National Sample Survey Office (NSSO). The NSO is the agency that actually produces the flagship products being exposed here: it runs the large household and enterprise surveys, computes the national income aggregates and compiles the price and production indices. The MCP-server announcement therefore sits inside MoSPI's broader, long-running push to modernise the delivery of official statistics — moving from static PDF releases toward APIs, open data and now AI-ready interfaces. A sibling release the same day described PAIMANA, MoSPI's web system for monitoring large central-sector infrastructure projects, underscoring that this is a ministry actively rebuilding its public-facing data plumbing.
For Prelims
- What it is: a Model Context Protocol (MCP) server launched by MoSPI that connects AI tools and platforms to verified official statistics held on the e-Sankhyiki National Statistical Portal.
- MCP — the protocol: an open standard interface that lets AI clients (assistants, apps, agents) discover and call data and tools published by an MCP server, replacing one-off custom integrations with a single common channel.
- e-Sankhyiki: MoSPI's National Statistical Portal for disseminating official datasets and time-series via structured access and APIs; the MCP server exposes this portal's data to AI.
- Coverage path: the Beta version carried 7 products; coverage has now been extended to 21 products on the e-Sankhyiki portal.
- Named products in the Beta set: Periodic Labour Force Survey (PLFS — employment/unemployment), Consumer Price Index (CPI — retail inflation), Annual Survey of Industries (ASI — the registered factory sector), Index of Industrial Production (IIP — short-term industrial output), and National Accounts Statistics (the GDP/national-income aggregates).
- Producing agency: these flagship products are compiled by the National Statistical Office (NSO) under MoSPI; the MCP server only re-serves them, it does not generate new statistics.
- Stated goal: machine-readable, AI-accessible official statistics, so AI systems answer using authoritative source data rather than scraped or stale figures.
- Where announced: a Rajya Sabha written reply on 23 March 2026 by MoS (I/C) Rao Inderjit Singh.
Know the five flagship products it exposes (a "match the pairs" set): PLFS → labour force, employment and unemployment indicators; CPI → retail (consumer-level) inflation and the basis for the inflation-targeting framework; ASI → the principal source on the organised/registered manufacturing sector; IIP → a monthly volume index of industrial output across mining, manufacturing and electricity; National Accounts Statistics → GDP, GVA and the national-income aggregates. Pairing each acronym to the right domain is exactly the kind of one-mark distinction UPSC builds questions around, and all of them sit under the NSO/MoSPI umbrella.
What this is NOT: the MCP server is not a new survey, a new index or a new statistical product — it produces no fresh data and changes no methodology; it is purely an access channel that re-serves data e-Sankhyiki already holds. It is not the e-Sankhyiki portal itself, but a protocol layer that sits on top of it. "MCP" here is the Model Context Protocol, a data-access standard for AI; it is not a hardware "server" in the everyday sense, nor should it be confused with unrelated abbreviations. And the underlying statistics are produced by the NSO under MoSPI — the MCP server does not transfer that statistical authority to any AI tool; the AI merely reads verified numbers, it does not certify them. Finally, this is distinct from MoSPI's PAIMANA system, which monitors infrastructure projects, not statistical dissemination.
Why it matters
The problem being addressed is the gap between where reliable public data lives and where people increasingly ask their questions. As more users turn to AI assistants for quick answers, those assistants often draw on whatever text they can find on the open web — which means official figures can arrive second-hand, out of date, partial or simply wrong, with no clear provenance. By publishing an MCP server over e-Sankhyiki, MoSPI lets an AI client fetch the verified, current number straight from the statistical system, with the source attached. That is a direct response to misinformation risk: it raises the chance that an AI-generated answer about Indian inflation, employment or industrial output rests on the authoritative dataset rather than on an approximate web echo.
It also fits a broader governance theme of data-driven, transparent administration. Open, machine-readable official statistics lower the cost for researchers, journalists, businesses and citizens to interrogate government data, and an AI-native interface lowers that cost further by letting a non-technical user pose a plain-language question and still reach a structured, sourced answer. For the state, the same channel can feed internal analytics and evidence-based policy. This is a concrete, citable instance of e-governance and the adoption of emerging technology inside the official statistical system — useful in answers on transparency, digital public infrastructure and the application of AI in everyday governance.
A balanced view carries the limits too. An access channel is only as good as the data behind it: if a product is delayed, revised or methodologically contested, the MCP server faithfully passes that on. Exposing data to AI tools does not by itself guarantee that the AI interprets it correctly — a model can still misread context, mix up reference periods or hallucinate around a correct figure — so the verified-source channel reduces, but does not eliminate, the risk of misleading AI answers. Coverage is also still partial at 21 of the portal's many products. The measure is therefore best read as an enabling step in modernising statistical dissemination, valuable as an example of AI-in-governance rather than as a finished solution.