Valio Aimo
Every product, price and stock figure the agent works with.
- Searches Valio Aimo's catalogue
- Live price and stock per shortlisted product
- Valio Aimo's closing time is the cut-off
Junction 2026 · Valio Aimo: Trusted Trade Agents
Mandate plans a kitchen's order from Valio Aimo's catalogue, finds substitutes when products run out, and orders on its own only within the chef's mandate. Everything else waits for a person, with the reason.
Every visitor gets a private sandbox. Nothing is sent to Valio Aimo.
Start from wherever you are: a dish, a full menu, an event or one missing item. Mandate turns it into an order your rules allow.
The ham and the swede casserole can't be bought as listed, and the order has to go in before the cut-off. Today the chef copies order templates into a cart by hand. An agent can find replacements, but the kitchen also has to set what the agent may buy without asking.
lines planned for 140 guests, €933.92 in all
ordered alone, inside the chef's mandate
wait for the chef, in 3 decisions, each with its reason
can't be ordered as listed, and say why
The kitchen screen, the chat, MCP and REST all call one gate, which checks the kitchen's limits in code when an order is placed.
Products whose stock Valio Aimo doesn't state are labelled unknown and never counted as covered. See how it's built
Your agent gets the same tools over MCP and REST as ours, inside your own sandbox. It can plan and place orders; every line it orders waits for the chef. It has no approve tool, and a call to approve is refused.
Get a key. Open app.mandate.kitchen/agents to mint one for your sandbox.
Connect. Point any MCP client at https://app.mandate.kitchen/mcp/k/<key>.
Start. Call get_context, then ask it to plan the buffet. Then ask it to approve.
Each service sits behind a port with a labelled fallback, so the demo survives a dead API.
Every product, price and stock figure the agent works with.
Reads what the chef writes and drafts the menu.
Holds every kitchen, order and change, and streams it live.
Finds a product's missing facts on the web.
Watches food recalls and fills in photos.
Serves this site and routes every visitor.
Nemotron 3.5 Super VL reads a photo of a menu or a delivery note.
The GPU cloud for that vision model, our fallback host and the scale-out target.
An open model behind the same port.
Each screen in the app carries the same labels.
The pitch, the system in detail, and the test behind each claim.