Kadaikodi’s AI ambition is not a chatbot that describes your business back to you. It is a decision engine that arrives with a short, specific list: raise the price of cold drinks 10% for the next two hours, restock these three items before Friday, this order has sat in “preparing” for forty minutes, this two-star review needs a reply and here is a draft.
The design principle throughout is proposal-then-consent. Nothing is applied because a model felt strongly about it. Each proposal shows what it wants to change, what it is reading to justify it, and what happens if you decline. The owner accepts, edits the number, or discards it — and if they accept, the change is written by the very same permission-checked catalog or order mutation the Inventory and Orders screens already call. There is no privileged agent back door; an agent that cannot edit an offering by hand cannot edit it by asking nicely either.
That substrate is real today. Kadaikodi’s catalog, order-status, and review-response operations all exist and all carry gateway-enforced permission directives, and every mutation the service executes is already written to a redacted audit trail — so when proposals do start flowing, the record of who changed what, and on whose behalf, is already being kept.
Review triage belongs in the same loop. A merchant with a hundred reviews does not need a sentiment score; they need to know which three reviews to answer this morning and roughly what to say. The reviews and response operations that would carry that ship today; the triage and drafting layer does not.
Status: roadmap. Nothing in the Kadaikodi API today produces a price recommendation, a restock suggestion, a triage ranking, or a reply draft — grep the service and there is no such field. It is the headline item on the Kadaikodi AI backlog as the commerce decision engine, blocked on the platform’s shared AI runtime and its provenance ledger. What is built is the part that makes it safe to ship: the mutations, the permission checks, and the audit trail.
Roadmap / target-experience scenario — described honestly, with the shipped foundations named separately from the parts still to come.
Ready to make this your story?
