JANSETU AI is a Digital Public Good that turns citizen voice (Gujarati, Hindi, English via voice/text/photo) into evidence-backed development priorities via deterministic priority scoring and human-governed review.

How does the priority scoring work?

Deterministic engine v1: demand 30% + infrastructure gap 20% + population impact 15% + vulnerability 15% + urgency 10% + feasibility 10%. Every component and weightVersion is persisted and auditable.

Is JANSETU AI a complaint portal?

No. It is a public intelligence layer that clusters citizen requests, fuses GIS/demographics/infrastructure data, and recommends candidate projects for human government review with full audit logging.

Process · Six steps · v1

How JANSETU works

Six steps from voice to measurable impact. Frontend is untrusted. Backend owns validation, clustering, scoring, audit. Gemini may understand and explain — never decide.

Raise a request Developer docs API reference

Citizen Voice

Speak in Gujarati, Hindi, English — voice, text, or photo. Original text is preserved verbatim; location is opt-in.

Evidence-led

AI Understanding

Gemini structures category, urgency, affected groups, location — validated server-side via Zod. Never rewrites meaning.

Evidence-led

Data Fusion

BigQuery GIS joins demographics, infrastructure indices, investment gaps. Evidence refs are persisted, never invented.

Evidence-led

Priority Intelligence

Deterministic v1: demand 30% + gap 20% + population 15% + vulnerability 15% + urgency 10% + feasibility 10%.

Evidence-led

Government Action

Policymaker reviews evidence, audit-logged approve/reject with editable reason. No auto-funding.

Evidence-led

Impact Tracking

Baseline → Target → Actual with observed vs modeled, measurement source & quality. Close the loop.

Evidence-led

DETERMINISTIC ENGINE V1

Every score is reproducible. Every weight is versioned.

Priority = demand 30% + infrastructure gap 20% + population 15% + vulnerability 15% + urgency 10% + feasibility 10%. Gemini explains drivers/limiters — never overrides.

Demand 30%92

Requests clustered

4,218 Vadodara

Gap 20%88

Infrastructure gap

Road 38/100

Population 15%80

People affected

12.4k density-aware

Vulnerability 15%82

SC/ST, women, children

Weighted equity

Urgency 10%90

Monsoon isolation

45→28 min

Feasibility 10%64

Terrain / cost / time

Phased viable

Band: critical ≥80 · high 65–79 · moderate 45–64 · low <45. Every component + weightVersion persisted for audit. See API.

AI governance

Gemini MAY understand, classify, translate, explain, draft.
MUST NOT invent evidence, alter weights, approve funding, override authoritative data.

TRY IT

See clustering live

Submit a Gujarati request and watch intake → cluster → score happen server-side.

Raise a Community Need

What happens after Submit?

Validate → save → media → Pub/Sub
Gemini intake → Zod → geography → cluster
Enrich demographics/infra/investment → deterministic score
Candidate project → Human review

Trusted by design

Never uses religion, caste, politics in scoring. Centroids only, never individual positions. Full audit log.

Privacy About governance