Sawa Sentinels

In Cameroon, when a school's water point fails or latrines overflow, reporting is chaos: phone calls, paper, silence. Rural schools wait weeks while loud downtown ones shout first. WASH Sentinel ends that. WASH Sentinel is a dashboardless AI platform for school WASH emergencies. There is nothing to learn: operators simply talk to Sentinel, an omnipresent voice agent that generates cards, charts, maps, briefings, and decision dossiers on d emand. A school director reports the way she actually communicates — a WhatsApp text, voice note, or breakdown photo, in French, English, Douala, or Ghomala'. Sentinel transcribes (Whisper + Meta Omnilingual ASR), triages with a priority score, and merges duplicates, so five teachers reporting one broken pump create one case, not five. Then it helps spend public money wisely: a repair-vs-replace copilot weighs repair ratio, breakdown frequency, and two-year savings; a territorial equity dispatcher ranks schools by need, so the quiet rural school outranks the loud downtown one. Every figure is human-verifiable on admin pages, because we never trust AI 100%. The loop closes automatically: each time the case moves assigned, in progress, repaired with its real cost, the director is notified on WhatsApp, in the language of the field. Local-first and sovereign (fully offline-capable), model-independent with runtime local-to-cloud switching from Settings, and 100% open source.