Team United

AI Market Assistant helps small traders in Nigerian markets get found by customers. THE PROBLEM The idea started in Kubwa market, where we spent a long time looking for one screw because nobody could tell us which shop sold it. Nigerian markets are mostly undigitalized: buyers can’t tell which shop has an item, prices vary with no visibility, and traders lose sales because customers can’t find them. Traders told us their biggest problem isn’t tracking profit, it’s finding customers. Most existing tools ask a busy trader to fill in forms, so they don’t get used. WHAT WE BUILT (live today) Traders log a sale or restock by speaking or typing a normal sentence. Meta’s Muse Voice Transcribe writes down the voice note, and Muse Spark turns it into a structured stock change (item, quantity, action). The trader reviews and confirms every entry before anything is saved, so the AI never changes stock on its own. Stock is derived from an append-only ledger of events, so history is never overwritten and mistakes are corrected with traceable adjustments. Concurrent sales of the last item are blocked, so a shop cannot sell what it doesn’t have. Buyers search without an account, pick the exact product and pack size, and see the nearest shops with estimated availability, a landmark and a map. Results update automatically within seconds when a shop sells out. Buyers can report “Found it” or “Not found”, which feeds back into how much the platform trusts a listing. Traders see low-stock and near-expiry alerts and a daily summary of recorded sales and stock. All data in the demo is simulated (30 fictional shops around Kubwa) and labelled as such. Trader and buyer data is kept separate, and buyer coordinates and raw search text are not stored. WHERE IT GOES NEXT (planned, not yet built) Phone notifications for traders: daily sales reports, expiry warnings and restock alerts delivered by WhatsApp. Demand insights: if many buyers in an area suddenly search for a product that a shop doesn’t stock, the trader is told to restock. This will be anonymized and shown only when enough different buyers are involved. Weekly and monthly sales ledgers, so traders have a record that can later support loan applications with a licensed financial partner. Support for Nigerian Pidgin, Yoruba, Igbo and Hausa, with testing before we claim it. Photo search, so a buyer can snap an item, and road routes to the shop. A pilot with real traders in Kubwa, starting with interviews and a small group using the app daily. Our approach is to make logging so easy that traders do it, because that fresh data is what makes shops discoverable. Inventory tracking is the cost of entry, and discoverability is the value.