Voice-first, multichannel dialogue
Producers ask a question by voice note or text. WhatsApp is the first target channel, alongside web, mobile and call-centre access.
The voice-first interface that makes agricultural and climate expertise accessible in the right language.
Producers ask a question by voice. WOURI understands the context, consults validated sources and returns a simple, traceable and adapted answer.
Information can be accurate without being directly useful. When it is late, too general, text-only or unavailable in the producer's language, it does not always lead to action.
The challenge combines language, context, speed, trust and traceability. An alert or recommendation loses much of its value without a date, an area or a route to a human expert.
WOURI targets that junction: connecting a field question to controlled sources, then returning an understandable answer without confusing the interface with scientific authority.
WOURI does not invent weather data or agronomic science. It orchestrates, explains and distributes useful information, then routes the request to a human when the situation requires it.
Producers ask a question by voice note or text. WhatsApp is the first target channel, alongside web, mobile and call-centre access.
WOURI identifies the language, crop, area, stage and intent so that each request is placed in its useful context.
Answers rely on controlled knowledge and dated data. The source, area and confidence level remain visible.
A critical, uncertain or insufficiently documented request is flagged and can be routed to an identified agent or expert.
WOURI receives a voice or text question, identifies its context, consults authorised sources and returns a simple answer in text or audio. Critical or uncertain cases are routed to a human.
Illustrative internal demonstration. Scientific, linguistic and field validation is the next step.
WOURI is designed to understand and return information in French and local languages. Dioula is being tested internally; other languages will progress with native speakers and gradual domain validation.