Nirvahan prototype

How it works

The problem, what changed, and how it would run at scale. Built for all of India.

See it as a citizen
dummy image

The problem: a vehicle owner in India cannot find out what they owe or what is about to expire, and cannot act on it, without knowing which of ~25 portals, which form, and which sequence. So they pay agents, pay twice, or get fined. Sarathi rates 1.7★ on Trustpilot; mParivahan 2.2★ on the App Store.

35 cr
registered vehicles · 20 cr licences
every one touches these flows
₹500 cr
unpaid fines in Karnataka alone
mostly, nobody knew
25
portals a citizen juggles today
→ one screen
4
separate logins for the same person
→ one, fixed via Aadhaar
2 cr
unpaid traffic notices in Karnataka
→ told before it matters
7 → 3
steps to renew a licence
→ no RTO visit
Today
VahanSarathieChallanNextGen eChallanVirtual CourtPUCNTRmobile update ×2mParivahanSlotsPaymentsVerify paymentState policeKarnataka OneDigiLockerFancy numbersPermitsFitnessDashboards ×4Grievance ×3

Each with its own login, captcha, state dropdown and vocabulary. The citizen is the integration layer.

Nirvahan
One Feed
Everything due · what happens if ignored · when · one button

Same data, same policy rails. A citizen-facing layer over them.

Time to get it done

Same six tasks, today vs Nirvahan. Log scale: each step right is 10× longer. Estimates from user guides, reviews and news; not a study.

Today Nirvahan
Find out if you have a challan
3–5 min
15 s
Pay a challan
5–10 min
30 s
Contest a wrong challan
60 days (court)
2 min
Know the licence renewal fee
~5 min
instant
Renew a licence
30–60 min
~2 min
Know if a transfer will go through
days (by rejection)
10 s
1 s1 min1 hr1 day1 mo

What changed: nine ideas

Each one removes a root cause, not a symptom.

01

Link once

Plate or licence, one OTP. Stale number? Verify with Aadhaar at the moment it fails. Fixed for every service.

02

The Feed

Home is a to-do list, not a menu. Each item: what · what happens if ignored · when · one button.

03

Intent search

Type what happened, or a form number. You never pick a form.

04

Case file

Every challan: photo, place, rule in plain words, and the options with their consequences.

05

Preflight

“Can you do this today?” before any renewal or transfer. Blockers with fix buttons, fee now vs later, RTO visit or not.

06

Zero-form

We ask only what the record can't tell us. At the end: “We filed Form 9 + 1A for you.”

07

Tracker

A submitted application is just a Feed item in its future state.

08

Explain-on-tap + a model

Four narrow jobs for an OpenAI model, grounded on a rules file. Never invents a fee.

09

Push

Every deadline becomes a message with a link. The Feed comes to you.

Built for all of India

Vahan, Sarathi and the national challan platform already cover every state. Where a state runs One Nation One Challan, Nirvahan's challan feed works on day one; elsewhere it reads the state police feed. Each state is one rules file and one challan feed; the UI never changes.

One Nation One Challan live (27) State police feed Demo citizens (Bengaluru, Karnataka)

Boundaries: Survey of India via the datameet open dataset; the J&K shape predates the 2019 J&K/Ladakh split and covers both. ONOC membership as listed on the eChallan portal (Aug 2026). Schematic, not to scale.

How it would run at scale

No new policy. A citizen-facing layer over systems that already exist.

INTEGRATIONCONSENTVVahan (RC)read + writeSSarathi (DL)read + writeeCeChallan / ONOCread + writeVCVirtual CourtreadPFState police feedsread + writeIIBIIB insurance · PUCreadBLBank lien registryreadNirvahanone FeedYouone login · OTP or Aadhaar

You log in once, to one screen. The seven systems around the edge are the government systems that already exist; none of them change. Nirvahan is the layer in between: every call passes the consent ring (Aadhaar e-KYC · OTP), reads flow in, writes flow out.

  • No new policy. Faceless Aadhaar services, ONOC, e-documents and bank-to-registry loan closure already exist.
  • Reads vs writes. Reads: RC/DL/challan/insurance/PUC. Writes: payments to the state gateway, applications to Vahan/Sarathi, grievances to the issuing police.
  • The model never touches money or law. Four jobs, grounded on rules.json; model gpt-4.1-mini; written fallbacks.
  • Privacy. Nothing leaves the device except the four model calls; no PII server-side; no cookies, no analytics.
  • Rollout. One state = one rules file + one feed. Karnataka first, because the demo citizens live there.
  • What we can't fix. Server capacity, SMS gateway lag, police data silos, RTO decisions. Shown honestly.
Proposed process changes (not current law)
  • A filed dispute should pause the 60-day Virtual Court clock. Today it doesn't; the prototype shows the clock still running.
  • Intimation of sale (Form 29) should auto-redirect later challans to the buyer.
  • Every issued challan should carry the camera frame and a plain-language reason by default.
Independent hackathon prototype. Not affiliated with MoRTH / Parivahan. All data, OTPs and payments are mocked.