AI2 min read
Managers can now ask Ronono questions in plain French or English. Every figure in the answer comes from the centre’s own records, and an answer that invents one is thrown away.
A collection centre answers the same questions every week. How much did this farmer bring? What does that reseller still owe? Who are the best suppliers this month? How much milk is in the tanks, and how much is expected tomorrow? Until now, each answer meant opening the right page and reading the right column. Now a manager can simply ask Ronono, in plain French or English, the way they would ask a colleague.
How an answer is made
The language model never invents the answer. It reads the question and chooses one approved, read-only query from a fixed list, together with its details: which farmer, which period. Ronono runs that query on the centre’s records, with the permissions of the person asking, so a clerk only gets answers about what a clerk can already see. The model then phrases the result as a sentence.
That sentence is checked before anyone reads it. Every number in it is compared with the figures the query returned. If the model states a number that is not in the data, its sentence is thrown away and Ronono shows a plain answer built directly from the figures instead. The same plain answer is used when the AI server cannot be reached, so the question is still answered.
What you can ask
Deliveries and quality for a farmer, farmer and reseller balances, the top farmers of a period, sales, expenses by category, stock on hand, upcoming payouts and collections, tomorrow’s volume forecast, today’s surplus and the production batches.
The model is an open model, hosted by Vayu Consulting, and a centre can point Ronono at its own AI server instead. AI calls are counted per month: 300 on Starter and 2 000 on Pro. The 30-day trial includes everything, so a centre can try the assistant on its own records from the first day.
