SiteProof
Case study 05: AI app · Field service
A technician speaks in Urdu, Hindi or Malayalam; the client gets a signed English and Arabic report.
AI agent
What the AI does: Turns a technician's voice note and photos into a signed English and Arabic report, each time one is recorded.

- signed job reports
- EN + ARsigned job reports
- Urdu, Hindi, Malayalam
- Voice inUrdu, Hindi, Malayalam
- data stays on the device
- No databasedata stays on the device
The problem
Technicians in the Gulf often speak Urdu, Hindi or Malayalam, but their clients expect a written job report in English and Arabic.
What I built
- 01The technician records a voice note and adds photos; the app returns a signed job report in English and Arabic
- 02Speech-to-text with Whisper, then a chain of language models with fallback
- 03Private by design: no database, everything stays on the device except the audio being transcribed
Stack
- Next.js
- TypeScript
- Groq Whisper
- Cerebras
- Gemini
- PWA

