Ask the AI Interface What Your Rules Are
See where you stand before deciding where to go.
The AI Distribution Interface has a new read-only tool that lets a tenant’s assistant see the tenant’s own supplier enablement, connection references and rules. It takes no tenant ID and reads only the caller’s own tenant; keys, credentials, commission and cost fields are not part of its response. The tool is off by default and currently open in the test environment.
“Why does my search return no hotels?” It is an unsurprising first question for a tenant that has just connected through an AI assistant. The answer usually sits in the setup rather than the search: which suppliers are enabled, whether a connection is usable, which rule filtered the results. Until now the assistant could only guess, or send someone to click through the portal page by page.
Letting an assistant read configuration starts with deciding what it reads, and whose. This tool accepts no tenant ID; the caller’s identity alone decides which tenant it reads. Keys, credentials, commission and cost fields are not part of the response: they are not read and then masked, they were never put in.
The read-only tool returns only the tenant’s own setup: the tenant itself, the supplier enablement and connection references stored on it (id, supplier, name, status), the bound rules, and each rule row’s name, status, condition and target, up to 100 rows with an explicit truncated flag beyond that. We deliberately do not fill gaps from a parent account: if nothing is attached to the tenant itself, the list is empty.
Calling it changes nothing. Where the tool is not enabled, a direct call returns a clear tool error rather than an empty result that looks like success. Once the picture is clear, enabling demo suppliers or changing a rule remains a separate step, taken by a person or by an operation that carries the matching permission.