Smart matching
AI scores, a person decides, then it reaches you. Every recommendation carries a match score and a reason written by a person.
Build the seller profile
Categories, markets, store size, fulfilment capability and past sell-through, recorded item by item.
AI shortlist
Profiles meet SKU market data; the model returns candidates with reasons.
Human review
A selection manager checks the reasoning, drops bad fits and adds what the model cannot see.
Adaptive push
Weights adjust with feedback and real sales — it gets better with use.
Why it has to be AI plus people
AI can score tens of thousands of products against thousands of sellers in seconds. People judge what the data cannot show: a seller's cash position, a local taboo around a product, whether a supplier really holds its lead times. With AI alone you would get a pile of plausible but wrong recommendations; with people alone there is no scale.
Commission applies to AI recommendations too: when an AI push closes a deal, the slot and the channel share in it.