ناطق (Naatiq)

A voice-first WhatsApp career agent for workers the CV economy leaves behind.

Naatiq turns a spoken WhatsApp conversation, in the user's own Arabic dialect, into a professional, bilingual (Arabic + English) PDF CV — no app to install, no forms, no literacy or typing barrier. It then offers job-category matching and mock-interview practice, all inside one WhatsApp thread.

See Wathefti_PRD.md for the full product spec.

Architecture

Layer	Choice	Role
Database	Supabase Postgres	users, profiles, conversations, matches, per-user state
File storage	Supabase Storage	voice-notes (inbound audio), cvs (generated PDFs)
Orchestration	Supabase Edge Functions	webhook + pipeline glue
LLM	Claude Sonnet	interviewer persona + CV content
Speech-to-text	Groq Whisper	Arabic voice → text
PDF rendering	Gotenberg / hosted API	bilingual RTL CV → PDF
Messaging	Twilio WhatsApp Sandbox	WhatsApp in/out
Dashboard	Cowork + Supabase	impact visualization

Supabase project ref pehlozswmyxocwznyjzp, region eu-west-2.

Repo layout

supabase/migrations/   SQL migrations (schema + dashboard views)
supabase/functions/    Edge Functions (webhook + pipeline)
cv/                    bilingual RTL HTML CV template + fonts
dashboard/             impact dashboard
docs/                  setup + notes

Status

• Phase 0 — Supabase project, schema, RLS, Storage buckets, dashboard views
• Phase 1 — WhatsApp webhook Edge Function (whatsapp-webhook, deployed)
• Phase 2 — Voice transcription (Groq Whisper) + Claude interviewer
• Phase 3 — Bilingual CV PDF generation (generate-cv, deployed)
• Phase 4 — Impact dashboard (dashboard/index.html, live via anon key)
• Phase 5 — Secondary features + polish
• Live test — set Edge Function secrets + wire Twilio sandbox, redeploy webhook

Setup

See docs/SETUP.md for the API keys and configuration required to run the pipeline. All secrets live in Supabase Edge Function secrets — never committed to git.