evidoria

← Back to browse

Good practice Imported

eSankhyiki MCP Server — India's National Statistics Office Opens Official Data to AI Tools via the Model Context Protocol

India · New Delhi · See the India profile · See the New Delhi profile

Evidence: Descriptive / self-reported Top 62% 53/100 · Ask Evidence Copilot about this practice

India's National Statistics Office launched a beta Model Context Protocol server on 6 February 2026, letting AI tools query official statistics directly. An independent test verified inflation and unemployment figures but found gaps in flagging methodology changes.

7
Datasets exposed at launch (6 Feb 2026)
21
Data products after expansion (within months of launch)
17
MoUs signed under Data Innovation Lab (by Jan 2026)
eSankhyiki MCP Server — India's National Statistics Office Opens Official Data to AI Tools via the Model Context Protocol

Details

Maturity
Pilot
Promoter
National Statistics Office (NSO), Ministry of Statistics and Programme Implementation (MoSPI)
Period
Beta launched 6 February 2026 with 7 datasets; expanded to 21 data products by mid-2026
Keywords
digital government, open data, official statistics, AI infrastructure

Context

On 6 February 2026, India's National Statistics Office launched a beta Model Context Protocol (MCP) server on its eSankhyiki portal, letting AI assistants query official statistics conversationally instead of via manual API navigation or file downloads. The source code is published on GitHub under an MIT licence.

Results

The beta initially exposed seven datasets and expanded to 21 data products within months. An independent technical blogger checked three specific queries against official sources -- December 2025 inflation figures, a Bihar unemployment rate, and a textile Wholesale Price Index comparison -- and found all three matched official releases exactly.

Conclusions

The same review flagged that the underlying LLM did not adequately warn about a Consumer Price Index base-year change affecting comparability, could not efficiently retrieve complete long historical series, and does not generate reproducible API calls a user could independently verify. As a beta pilot covering a fraction of India's statistical output, whether its simple-lookup accuracy holds as coverage and query complexity grow remains an open question.

Implementation
Indicative cost
Low (< €50k)
Time to results
Short (< 1 year)
Staffing & skills
NSO/MoSPI Data Innovation Lab team

Conditions for success

  • Open protocol (MCP) with open-source, MIT-licensed implementation
  • Existing structured official statistics APIs to expose

Common failure modes

  • No systematic accuracy evaluation, only an informal 3-query spot-check
  • LLM did not flag a CPI base-year comparability change

Commonly funded by

National / regional programmes

Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Attachments

Similar practices you may find useful