Haq
A voice-first welfare-benefits guide for construction workers across four Indian states, designed for low-literacy use and multilingual access.
Problem
Workers need understandable guidance on changing welfare rules without an app pretending to make a legal eligibility decision or connect to a government system.
How I solved it
- 01
Evaluates every eligibility result through deterministic rules over worker-supplied information, keeping the AI model out of the decision path.
- 02
Uses AI only to explain an already-computed result or answer an explicit free-text question with cited government sources in the worker's language.
- 03
Keeps profiles, work history, crew records and attestations in browser localStorage; the build uses mock or synthetic data and has no live government integration.
My role / what I owned
Product engineer and builder
- Built the deterministic eligibility, work-history and guidance flows for four state contexts.
- Implemented eight-language access with static and runtime speech support.
- Added the cited-answer and explanation boundaries without allowing AI output to decide a worker's eligibility.
How it was built
- Team
- A self-directed project in Nivish's private repository.
- Tools
- Next.js
- TypeScript
- Vitest
- OpenAI API
- Sarvam speech
- localStorage
- Use of AI
- AI explains a deterministic verdict or answers an explicit question; it never decides eligibility, and cited government sources remain the answer boundary.
Source and current state
- The private repository has more than 100 commits since the 10 August 2026 evidence review.
- Its current README documents 204 Vitest tests across eligibility, i18n, store logic and AI-boundary regressions.
- The audited source documents eight supported languages, local-only worker data and no live government-system integration.
- Status
- Private build with a public Vercel demonstration. The source repository remains private.
- What it shows
- The audited repository demonstrates the deterministic eligibility, multilingual guidance and voice-accessibility boundaries without making a government endorsement or integration claim.