Primary user
A PM-KISAN beneficiary who may use Hindi, an older phone or a slow connection and cannot easily interpret technical statuses.
Project submission
From payment status to one clear next step.
Paisa Kahan Hai is a bilingual, privacy-safe prototype for PM-KISAN beneficiaries with limited digital confidence—and the people who help them.
Open the working product No login, API key or personal data required.
Project summary
A safer interpretation layer for a difficult public-service journey
PM-KISAN gives eligible farmer families ₹6,000 a year in three payments. If one does not arrive, the government's Know Your Status service requires a registration number, CAPTCHA and one-time password (OTP). The Kisan-eMitra AI chatbot answers questions about payments, registration and eligibility in several languages.
Paisa Kahan Hai adds a fixed workflow: it checks possible problems in a documented order and presents one first action.
It is an independent Hindi and English guide for farmers who are not comfortable with technology, and for the people helping them. It does not access government records or collect personal details. Users can try a fictional example or select status messages they already have.
The result gives the cause, office, items to carry, words to say, checked costs or timelines, verification and escalation. When a visit is needed, an identifier-free office note separates what to take from what to request there and what to confirm before leaving. The full guide can be read aloud, printed, saved as a PDF or shared. A WhatsApp-style preview offers the same checked journey one question at a time.
The latest PM-KISAN payment reached more than 9.44 crore farmers. Official data also records lakhs of stopped payments linked to land records, Aadhaar-bank linking and e-KYC. These figures show the problem's scale, not people helped by this prototype. All examples are fictional, and connection to government records has not been built.
Helping user
A family or community “phone person” helping the beneficiary decide what office to visit, what to carry and what to say.
Demo video
Round 2: the problem, working journey and one differentiating feature
Feature focus: the office-ready first action. It turns a technical status into preparation before a visit, a request for the official or bank counter, and confirmations to obtain before leaving.
Previous demo — Round 2 replacement pendingThe new two-minute video will spend its second minute on the office-ready first action.
Open video directlyPerformance evidence
Measured on mobile—with the date and limit visible
A dated Lighthouse capture records the earlier live release quality, while a fresh custom Slow-3G audit checks the current public Round 2 release under a stricter connection profile.
100Lighthouse performance · 26 Aug
100Lighthouse accessibility · 26 Aug
0.75 sCurrent diagnosis readable on Slow 3G
2.40 sCurrent document ready on Slow 3G
Lighthouse mobile1.6 s
Largest Contentful Paint
Lighthouse mobile40 ms
Total Blocking Time
Current first load189 KB
189,113 bytes across 12 requests
Method: Lighthouse 13.4.1 used the live production URL and its mobile profile on 26 August 2026. The current public release was measured on 5 September 2026 using 400 ms latency, 400 Kbps down/up and 4× CPU slowdown, with cache disabled.
Open the complete Lighthouse report ↗Honest limit: the current 189 KB first load is above this project's internal target of under 100 KB. The target remains recorded as not met.
Scale and potential impact
A large public-service problem, without an inflated reach claim
Official 2026 figures show both the scale of PM-KISAN and the number of records affected by mandatory-condition stoppages that this product is designed to explain.
- 9.44+ crore
- eligible farmers received the 23rd instalment
- Scheme-scale context
- 16.52 lakh
- payments stopped under land seeding
- As of 16 March 2026
- 23.54 lakh
- payments stopped under Aadhaar–bank seeding
- As of 16 March 2026
- 32.95 lakh
- payments stopped under e-KYC
- As of 16 March 2026
23rd-instalment release — PIB / Ministry of AgricultureRajya Sabha Question 3226 — stopped-payment conditions
These are problem-space figures, not people reached or outcomes created by Paisa Kahan Hai. The three stoppage categories may overlap and must not be added together as unique beneficiaries. Product reach and comprehension remain unmeasured.
Product / build
A working rules product, not an AI-generated answer
Choose a safe route
Try a fictional case, select statuses already known, or get help reaching the official status channel.
Interpret deterministically
An ordered rules engine identifies the first blocker and sequences any later blockers.
Act with confidence
The result gives one first action, the right office, documents, a counter script and escalation.
- 23
- fictional scenarios
- 20,736
- guided combinations tested
- 2
- human-reviewed languages
- 0
- personal identifiers collected
Technical status language
One plain-language cause and one first action
Uncertainty at the service counter
An office visit note for preparation, handoff and confirmation
Hindi or assisted use
Human-reviewed Hindi, read-aloud and print
Older phones and slow data
Server-rendered results, system fonts and 320 px support
Family or community help
A shareable one-page guide and device-local checklist
High privacy consequence
No identifier, login, tracking or server-stored progress
Codex supported requirements, evidence reconciliation, architecture, implementation, bilingual review batches and QA. The builder directed the product and approved the Hindi; the rules engine—not an AI model—decides every result.
Evidence and validation
The problem structure and product facts were triangulated
- Public journey inspectionThe official Know Your Status entry gate was inspected without entering or submitting personal data.
- Official process guidancePayment, KYC and NPCI mapper guidance established that different blockers route people to different institutions.
- Official scale evidencePIB and parliamentary records establish scheme scale and separate stopped-payment categories, with overlap and reach caveats kept visible.
- Claim-level verificationA 72-claim workbook tracks sources and wording; 49 claims are verified, and every workbook-backed claim reachable in the product is primary-sourced and verified.
This validates the problem structure and product facts—not beneficiary comprehension or outcomes. No PM-KISAN beneficiary or frontline-worker study has been conducted.
End-to-end service thinking
The interface is one layer of a production service
This is a proposed safe path, not a claim that these integrations are available. The current prototype remains entirely outside government systems.
- 01Citizen or helper channelSelf-service, CSC or assisted use
- 02Consent and authenticationPurpose shown before access
- 03Minimal read-only adapterOnly statuses needed for guidance
- 04Authorised status providersPM-KISAN, payment and state systems
- 05Versioned rules and guideTraceable bilingual action output
Backend
Minimal status contracts, versioned rules, source traceability, safe fallbacks and correction handling.
Infrastructure
Encryption, strong authorisation, rate limits, audit logs, abuse monitoring, uptime and incident response.
Process ownership
Taxonomy maintenance, bank/CSC/agriculture-office routing, escalation, human support and service accountability.
Required dependencies are not confirmed capabilities: authorised data contracts, consent and identity standards, responsible system owners, operational support and a correction pathway would all need agreement before real-record use.
Transparency
What is real, what is simulated and what production would require
Works now
- Bilingual status interpretation
- Read-aloud, full-guide sharing and one-page print
- Conditional, identifier-free office visit notes
- Local-only progress tracking
- Mobile, keyboard and no-JavaScript paths
Simulated
- All example records are fictional
- Guided inputs are user-selected status labels
- No example represents a real beneficiary
Not built
- No government, bank or grievance integration
- No login, database, analytics or SMS
- No real-record lookup or grievance filing
External dependencies
- Authorised, documented data contracts
- Explicit consent and minimal data
- Authentication, rate limits and audit trails
- Abuse monitoring and human support
Not yet validated
- No beneficiary comprehension study
- No frontline-worker workflow study
- No measured reach or outcome improvement
- No physical screen-reader or Android test