Answers the call, records audio, and returns an SMS confirmation.
The exchange that answers the phone .
ShatrixVoice Mesh lets a farmer with a basic feature phone list produce and find buyers by speaking — in Hausa, Yoruba, Igbo, or Pidgin. No smartphone, no data plan, no app to install.
Most agricultural trade still happens without the internet.
Digital procurement platforms are built for people with smartphones and data. The farmers producing the food are often the ones who cannot reach them.
- Listings that require an app to view or create
- Forms designed for keyboards, not for people who work by phone
- Support and market information offered only in English
of rural smallholder farmers are effectively outside digital marketplaces
Any solution that assumes a smartphone excludes the majority of the supply base.A phone call, and the listing exists.
No app, no login screen, no form. The farmer calls a number and talks naturally.
- 01
The farmer calls
A local number answers in the farmer’s language. No data connection is needed at any point.
- 02
They speak the offer
Commodity, quantity, location, and timing — said the way the farmer would say it to a neighbour.
- 03
It becomes structured supply
The recording is transcribed and read into the same structured fields buyers already search.
- 04
Everyone is notified
The farmer receives an SMS confirmation, and matched buyers are alerted to the new supply.
Every field is read back to the caller for confirmation before it is published.
If a call drops or fails, the farmer still gets a summary and can reply by text.
Calls stay short and structured so the feature remains affordable at scale.
Four steps from a spoken sentence to a live listing.
The voice service is a thin, stateless layer. Shatrix Agro remains the system of record for every listing, supplier, and match.
Turns dialect and Pidgin speech into text, including code-switching mid-sentence.
Extracts the commodity, quantity, location, and timing as structured data.
Creates a pending listing, runs matching, and notifies the right buyers.
Accuracy on Nigerian languages is the deciding factor, not the infrastructure. We benchmark recognition quality on real recordings before committing to a provider.
How we plan to get there.
Each stage is measured by whether a real farmer can complete a real trade, not by whether the technology works in a demo.
-
Phase 1
Next
Recognition benchmark and call prototype
Measure transcription accuracy across Hausa, Yoruba, Igbo, and Pidgin on real recordings, then prove one commodity listing end to end.
-
Phase 2
Planned
Pilot with one cooperative
Voice listings, SMS confirmations, and buyer alerts, live for a single region and reviewed with the farmers who used it.
-
Phase 3
Exploring
Multi-language rollout
Expand dialects, add outbound calls to buyers, and connect voice listings to the wider procurement network.
Help shape how this is built.
If you work with farmers, cooperatives, or agricultural finance, we would like to hear what your suppliers actually need.
- Early access to the pilot
- Input on the languages and dialects we support first
- Updates as each milestone is reached