AI project · Civic tech · Live

Responsible Nagrik

Apni colony ki awaaz bano.Be the voice of your neighbourhood.

A mobile-first platform where Indian citizens report local problems like potholes, open drains and uncollected garbage in Hindi, English or Hinglish, and see how their whole neighbourhood is doing.

psst, the language switcher further down actually works ↓

RoleFounder, PM, designer, builder
Built withLovable
Regions7 across India
StatusLive at responsiblenagrik.com
Report screen with 16 issue types
Home screen with language switcher and one-tap issue types
Quick select chips that fill in the complaint description
ProblemComplaints vanish

Civic complaints go into a portal or a WhatsApp group and nobody sees what happens next.

BetMake it visible

If a locality's backlog is public, single complaints add up to collective pressure.

BuiltReport in 6 taps

3 languages, 16 issue types, guided photos, and a live status page for every locality.

MeasureFixes confirmed

North star: problems citizens confirm as resolved each week.

01The problem

You can report a pothole in five places and still never hear back.

An open drain outside your gate is everyone's problem and nobody's job. Here's what reporting one usually looks like today.

Day 1Post in the colony WhatsApp group

Ten people reply "same here." Nobody files anything.

Talk, no action
Day 3Find the right portal

Municipal corporation, jal board or electricity board? Each has its own channel.

Which department?
Day 3Fill an English-first form

Ward number, zone and category codes the citizen doesn't know.

Language barrier
Day 10Tag officials on social media

Works for people with followers. Most people don't have them.

Unequal voice
Day 30Silence

No status, no proof anyone saw it. The citizen stops reporting.

Language

Most civic forms assume English. Most residents think and speak in Hindi or Hinglish.

No closed loop

A reference number with no visible progress feels the same as no reference number.

Isolated voices

Fifty people complaining separately about one pothole looks like fifty small problems, not one big one.

No locality view

Residents can't see whether their area is improving or being ignored compared with the next one.

02Who it's for

Who I had in mind

S
Sunita, 46
Primary · resident
Context
Homemaker in Bhopal. Uses WhatsApp and YouTube daily, never a government portal.
Needs
Report in Hindi, in under a minute, without knowing which department owns the drain.
Shikayat toh kar doon, par pata kaise chalega ki kisi ne dekha bhi?
R
Rahul, 31
Power user · RWA secretary
Context
Runs the residents' association for 300 homes in Noida. Juggles WhatsApp, spreadsheets and ward meetings.
Needs
One list of everything pending in the colony, with photos, to take to the councillor.
I need evidence, not forwards.
W
The ward office
Future · accountability side
Context
Councillors, local journalists and civic groups who want to know where problems pile up.
Needs
A trustworthy, locality-level picture instead of scattered complaints.
Show me which streets need attention first.

Personas are illustrative.

The insight
When people report something and nothing visibly changes, they stop reporting. So I made the changes visible.
The product bet

Give every locality a public scoreboard. When pending, resolved and stuck issues are visible by PIN code, each report becomes part of a collective signal, and residents come back to check on their neighbourhood, not just their own ticket.

03Design decisions

What I designed, screen by screen

Responsible Nagrik home screen 1 2 3 4
1
Honesty before authority

The first thing on every screen says this is not a government site. It would be easy to look official. Being clear about what the product is builds more lasting trust, and keeps it on the right side of impersonation.

2
Language comes first

Hindi, English or Hinglish is chosen at onboarding before anything else, and can be switched from any screen. Hinglish isn't a translation. It's how most users actually type.

3
One big action

"New Complaint" is the only primary button. Checking on an old complaint is a quieter second card, because new reports are what feed the whole system.

4
Tap, don't type

The eight most common problems are shortcuts right on the home screen, in everyday words: Open drain, Pothole, Garbage not picked.

5
Plain-language categories

"Garbage not picked," not "Solid waste management." Labels follow the selected language, and in Hinglish mode each type shows both the English term and the everyday word ("Pothole / Gadha").

6
Sixteen types, not a free-for-all

Enough to cover real life, from stray animals to dengue mosquitoes, while keeping the data structured enough to compare localities. "Other issue" catches the rest.

7
Photos that get things fixed

The photo step asks for three specific shots: a wide shot, a close-up and a landmark. A field worker can find the problem and judge how bad it is. Photos are optional, so a missing camera never kills a report.

8
Quick-pick descriptions

"Quick select" chips fill in the description with one tap. Typing on a phone keyboard is where most forms lose people.

Issue type picker with 16 bilingual categories 5 6
Try it

Same screen, three languages

This is how the app's issue picker behaves. People pick a language at onboarding and can switch from any screen.

04The reporting flow

Reporting takes about a minute

I gave each step a time budget. If a step can't fit its budget, it gets cut or made optional.

1TypePick from 16 bilingual issue types~5s
2PhotosWide, close-up, landmark. Optional.~15s
3DescribeQuick-pick chips or free text~10s
4LocationGPS, search or manual address~15s
5CheckReview before sending~10s
6SubmitGet a reference ID to track~5s
0sTarget: under 60 seconds60s
05Area status

My favourite part: seeing your whole area

Every locality gets a live status card, refreshed every four hours and searchable by PIN code or city. It answers the question residents actually have: is anyone fixing things here?

Pending complaints

How big the backlog is right now.

Resolved this week

Shows momentum, so improvement is visible early.

% work complete

Resolved as a share of everything reported.

Stuck for a long time

Ageing issues, which carry the most frustration.

Comparison, designed carefully. Areas can be compared three ways:

Needs attentionGood workBy problem

I chose "needs attention" over "worst areas," and gave "good work" equal billing. The goal is to motivate whoever can fix things, not shame a neighbourhood.

06Where AI fits

How AI fits in, now and next

How AI built it

  • I built it end to end with Lovable, an AI app builder.
  • I wrote product requirements as prompts, checked every generated screen against the personas, and kept iterating until each flow felt right.
  • Owning the build meant every trade-off was mine to make, which is the part I enjoyed most.

AI roadmap (not built yet)

Vision
Photo to category and severitySnap a photo and the model suggests "open drain, high severity." Fewer taps, cleaner data.
LLM
Hinglish voice and text understanding"Gali ke kone pe 3 din se kuda pada hai" becomes type, location hint and duration.
Clustering
Duplicate detectionTen reports of one pothole merge into one issue with ten voices. That turns noise into pressure.
Routing
Right department, automaticallyMap issue type plus PIN code to the responsible civic body, so citizens never have to know.
Summaries
Weekly locality digestAn auto-written, shareable WhatsApp summary for RWAs: what's pending, what got fixed, what's stuck.
07Success metrics

What I'm measuring

These are targets for the first 90 days, not results.

North starCitizen-confirmed resolutions per week

Problems a resident marks as fixed. It only grows if people report, come back, and things actually get done.

  • Pilot localities 25 / week
Input metricsIs reporting easy?
  • Start-to-submit completion > 70%
  • Median time to report < 60s
  • Reports with photos > 60%
  • Reports with precise location > 85%
  • Reporters returning to check status, D7 > 40%
GuardrailsIs it trustworthy?
  • Spam or fake report rate < 3%
  • Duplicate reports per issue tracked
  • False "resolved" updates < 5%
  • Photos with faces or plates blurred
  • Status data freshness ≤ 4h
08Tradeoffs

Trade-offs I made

DecisionWhat I gave upWhy it was worth it
Independent community platform, not a government channelDirect routing to officials on day oneLaunch without permission, earn trust through transparency, and integrate with civic bodies once there's data worth sharing.
Photos optionalEvidence on every reportA report without a photo beats no report. Guided shots raise quality for people who do add them.
Fixed categories instead of free textSome nuance in what people describeStructured data makes locality comparison possible, which is the core of the product.
"Needs attention" instead of "worst"A punchier, more viral rankingShame creates defensiveness in the people who can fix things. Recognition creates competition.
Citizens confirm resolutionAuthoritative, verified closureNo official API exists. The person who reported it is the best judge of whether it's actually fixed.
09Go to market

Every colony already has a WhatsApp group. That's where this spreads.

Every RWA already has a group where complaints get discussed and forgotten. Responsible Nagrik gives that group a place to put them and a card to share back.

Resident reports a problem→ Shares the reference link in the colony group→ Neighbours add photos and reports→ Area status card shared weekly→ More residents report ↺
11What I learned

What I learned

Visibility is the feature

The form gets people in. Seeing their area's status is what brings them back.

Language is a product decision

Supporting Hinglish touched the copy, the categories and the default flow. It was much more than a translation file.

Finish the boring parts

A half-translated screen breaks trust fast. Area Status still shows Hindi in English mode, and it's first on my fix list.