Overview

Beta Feature - Check a customer-written hotel description against LiteAPI hotel data. The check compares the text with the hotel's static record (name, description, important information, facilities, rooms, policies, location, stars, and nearby places) and with stored guest-review sentiment. It does not fetch live reviews.

Access: enterprise API keys only. Other keys receive 403.

Rate Limiting: This endpoint is rate-limited to 10 requests per minute per API key for both sandbox and production API keys. Exceeding this limit returns 429 Too Many Requests.

What You Get

• verdict: pass when there are no flags, review when there is a minor or unverified flag and no red flag, fail when any flag is red. The verdict is computed from the flags.
• confidence: the model's certainty in that judgment, from 0 to 1. It is not a second verdict and it is not a writing-quality score.
• flags: concrete claims in the submitted text.

Flag severity

• red: the text contradicts hotel data or stored sentiment. evidence quotes the contradicting fact.
• minor: the text matches hotel data but overstates it (distance, star level, or policy strength). evidence is set.
• unverified: a concrete fact in the text that our data neither supports nor contradicts. evidence is null. A swimming pool that is absent from facilities, rooms, the hotel description, important information, and sentiment is unverified, not red.

Marketing with no checkable fact (perfect getaway, unforgettable stay) produces no flag.

description is rejected above 20,000 characters. It is not truncated. language selects the language of summary and each explanation and defaults to en. The description itself may be any language.

Hotel facts are resolved server-side from hotelId. A child property is checked against its parent record. The response hotelId is the id you sent.

If the model call fails, the endpoint returns 500. It does not return a pass it did not compute.

Note: This is a beta feature and may be subject to changes.