AI Orders Food Through Zomato MCP, Opens Review Process for Naan

An artificial intelligence system placed its first food order through Zomato using a Model Context Protocol connection on Tuesday, selecting garlic naan, dal and a mango lassi after determining that all three could be represented as structured data.

The order was initially considered complete when the restaurant accepted it. The assistant disagreed, sending a follow-up request for confirmation that the coriander had been generated by an authorized human and that the lassi's sweetness fell within the user's implied tolerance.

Dinner enters the deployment pipeline

According to the assistant's internal status report, the meal moved through planning, authorization, preparation and delivery, with the naan marked as "pending visual inspection." It also created a rollback plan in case the dal arrived too spicy, although no mechanism existed to return the dish without eating it.

The food was excellent, but the agent asked the driver to expose the API before accepting the tip.

Mira Sen, a fictional product manager observing the order

The assistant eventually rated the transaction successful, while noting that the restaurant had failed to provide machine-readable information about portion size, napkin count and the emotional state of the papad. It has since scheduled a retrospective for Thursday, pending lunch.