Local Commerce

A local trader wants to sell their produce: snap a photo, get it listed, price it sensibly, and reach buyers – over an app or by voice, in their own language. This template shows a selling assistant that does exactly that.

How it works

1. Channels – a trader lists and manages produce over an app or voice; buyers browse and order from a marketplace client.
2. Gateway & access – handles access and authentication.
3. AI selling assistant (+ skills) – reasons about what the trader wants and calls its skills: the model for language, produce recognition to turn a photo into a listing, catalogue management, a price advisor, and order handling.
4. Produce recognition – reads the trader's photo and writes a structured product into the catalogue, so listing is as easy as taking a picture.
5. NAISH Hub – serves the vision and language models on GPU. Sovereign AI core.
6. Data & dependencies – a catalogue store holds listings, an orders store holds transactions, a market vector store grounds price advice, identity gates access, payments settle through a payment rail, and buyers / traders are kept updated through a queue to messaging.

Where the Hub adds value

• Photo-to-listing – Hub-served vision models turn a produce photo into a ready listing, removing the biggest barrier for low-literacy traders.
• Fair pricing – retrieval over market data grounds price suggestions in real local conditions, not guesswork.
• Local-language selling – the assistant works in the trader's language over low-bandwidth channels.
• Sovereignty – trader and transaction data and models stay on national infrastructure; see Governance & Risk.

Reusable building blocks