ವಿVicharane

Track what you have filed

ವಿಚಾರಣೆVicharane

Bengaluru · Civic complaints & Right to Information

Say it in your own words. We'll say it in theirs.

Describe a civic problem the way you'd describe it to a neighbour, in English or Kannada. Vicharane works out who is responsible, writes it the way they are obliged to answer, and holds them to every deadline the law gives them.

Scroll down for product architecture & tech stack

How it works

01

Describe what went wrong

In your own words, English or Kannada. No forms, no legal phrasing, no need to know which department it belongs to.

02

We find who owns it

Bengaluru has five city corporations and 369 wards since the GBA reorganisation, plus BWSSB, BESCOM and others. We identify the right body, the right ward, and the officer actually responsible — with their phone number.

03

We write it so it has to be answered

A civic complaint first, phrased so it cannot be dismissed as vague and so it commits the department to producing proof of the work. If that fails, the same grievance becomes a Right to Information application asking for the records — because the law obliges an authority to hand over records, not to explain itself.

04

We watch the clock

Thirty days for a reply, thirty to appeal, forty-five for that appeal to be decided, ninety for the next. Miss one and you lose the right. We track all of them and draft the next step when it comes due.

The system assumes you'll give up

Most complaints are not ignored. They are marked resolved. When a department closes your complaint without doing the work, Vicharane turns it into a legal demand for the action-taken report, the work order, the completion certificate and the site photograph — the records that cannot exist if nothing was done.

Architecture & Platform Flow

A comprehensive view of how Vicharane transforms unstructured citizen grievances into legally binding, trackable resolutions using AI and modern web infrastructure.

1. The Resolution Pipeline

1. Citizen Input

Users submit their grievances in plain, unstructured language.

Native Kannada support
Accepts fully local inputs
Natural English support
Conversational descriptions
Zero friction
No legal knowledge required

2. AI Intelligence

Multi-stage LLM processing of the unstructured input.

Stakeholder Mapping
Identifies exact civic authority (e.g. BBMP)
Location Mapping
Extracts ward identities & coordinates
Problem Drafting
Converts emotion into actionable summary
Legal Translation
Formats into RTI Act 2005 structures

3. Statutory Drafts

Outputs strictly formatted legal documents ready for submission.

Section 6(1)
Original RTI Applications
Section 19(1)
First Appeals generation
Section 19(3)
Second Appeals generation
Bilingual output
Produces dual-language formats

4. Action Timeline

Dynamic dashboard tracking progress and prompting user action.

Submission Log
History of all filed documents
Statutory Deadlines
Enforces the 30/45/90 day RTI clock
Next Actions
Intelligent follow-up prompts

2. Technology Stack

Client Application

  • Framework: React 19 + TanStack Start (SSR)
  • Styling: Tailwind CSS + Radix UI Primitives
  • Visuals: Three.js (3D Ward Maps)
  • Routing: TanStack Router (Type-safe)

AI Infrastructure

  • Gateway: Lovable AI Gateway / Groq API
  • Primary Model: GPT-4.1 class (openai/gpt-5.6-sol)
  • Fallback Models: Moonshot MoE & GPT-OSS-120b
  • Strategy: System prompts mapped to Karnataka legal limits

Backend & Data

  • Database: Supabase (PostgreSQL)
  • Auth: Supabase Auth & RLS Policies
  • Hosting: Vercel Edge Serverless
  • Validation: Zod (Strict schema typing)