Content Enrichment
Content Enrichment API
PageMore page actions
The Content Enrichment API delivers rich, ready-to-use hotel content – structured details, localized marketing texts, imagery, and more – through a single, consistent interface. Your hotels arrive already enriched and accurate, so you can put high-quality content in front of travelers without building and maintaining your own content operation.
What You Can Do
The API gives you four things: read your enriched hotels, manage the templates that shape your marketing texts, generate that text on demand, and run our image analysis over imagery of your own.
Get Enriched Hotels
Retrieve your enriched hotel content whenever you need it. Fetch a single hotel with getEnrichedHotel, page through your full set with listEnrichedHotels, and each hotel arrives ready to render, with its structured details, localized texts, and imagery.
Identify hotels your way
- Request side: look a hotel up by the
ExternalHotelIDselector usingproductid(the hotel's id in your own product system, the identifier you use internally),giataorpeakworkgoalid(the hotel's Peakwork Goal ID, a positive integer as a string), with no pre-mapping to our ids needed. Several keys in one selector must all match the same hotel, and a value that is not a positive integer forgiataorpeakworkgoalidreturns 400. The Vision AI ingest endpoint does not takepeakworkgoalid. Our direct selectorsHotelIDandGlobalHotelIDstay available whenever you already hold them. - Fetch your whole set at once: both list endpoints accept an
ExternalHotelIDsfilter, a list of selector groups that is resolved to the matching hotels, so you can pull exactly your set of hotels by your own codes in one call. A hotel that is unknown or not in your project returns 404 ongetEnrichedHoteland is simply absent from a list. - Response side: every hotel carries our identifiers
IDandGlobalHotelIDplus theExternalHotelIDsmap of external ids as strings, today"ProductID"(the hotel's id in your own product system, the identifier you use internally, stored per project when a mapping exists),"Giata"(when the hotel has one) and"PeakworkGoalID"(when the hotel has one). The map is extensible, so read the keys you need and ignore the rest.
Query by ExternalHotelID and receive ExternalHotelIDs, the same concept in both directions.
Use getContentSettings to read your project's content configuration, such as the locales available to you, so your integration always knows what to expect.
Manage Text Templates
Your marketing texts are produced from templates that you control. Use listTextTemplates to see your current templates and saveTextTemplate to create or update one. Templates let you shape tone, structure, and emphasis so generated descriptions match your brand and surface the details that matter to your audience.
Analyze Your Own Images
Send us image URLs and receive the same analysis our enriched galleries carry, including accessibility alt text in your locales, without ingesting a hotel catalogue first. See the Vision AI page for the submit and poll flow.
Generate Text
Produce localized marketing text from your templates in two ways.
- Generate asynchronously – Call
generateTextto start a generation job. It returns immediately with an id, and you pollgetTextGenerationStatuswith that id until the job is done. This pattern keeps long-running generation off your request path and scales cleanly across many hotels. - Preview synchronously – Call
previewTextto generate and return text in a single request. Use it to review the output of a template before you commit to a full generation run.
How Content Enrichment Works
Behind each hotel is a body of work you do not have to build or maintain. Here is what shapes the content you receive.
Verified, reconciled facts. Hotel details are gathered from multiple independent sources and reconciled with trust weighting, so conflicting claims are resolved instead of copied through. The result is accuracy you can put in front of travelers.
Star grades with provenance. A hotel can hold several star grades at once, and StarsMap serves them side by side, one entry per source, so you can pick the grade you trust. The sources are your own project grade (project, present only when you set one), the official classification in the hotel's country (national), a classification association's grade (association), and the grade the hotel states on its own website (canonical). The default entry is always present and is the first of these in that order, falling back to our collected rating, and the top-level Stars always equals it. Each entry also flags whether the grade is a Superior class and whether the hotel is classified as a Garni.
Guest ratings in one score. GuestRating tells you how guests rate a hotel and what they say about it. ReviewScore is one overall score from 0 to 100, aggregated from the guest review scores we hold for the hotel, each calibrated for how its review platform usually rates. SubScores adds a score from 0 to 100 for Cleanliness, Service, Location, Rooms, Food, Value and Facilities, showing how guests rate that aspect compared with the hotel overall. A category without data is left out, never sent as 0. Texts holds, per locale, a short AI-generated Summary of guest opinion with up to five Pros and up to three Cons; it quotes no individual review. A hotel without a rating has no GuestRating.
Understood imagery. Every image is analyzed and categorized (pool, room, dining, and more), tagged with the features it visibly shows, scored for appeal, de-duplicated, and captioned. Galleries arrive curated and orderable rather than as a raw photo dump, and the same analysis is available on demand for imagery of your own through the Vision AI endpoints.
Structured, normalized attributes. Facts land in a consistent ontology with stable keys and industry GlobalTypes codes (GTCodes), including free-of-charge qualifiers (IsFree) and, where the sources say so, whether a facility or activity is on the hotel's premises or in the surrounding area (IsOnSite), so your integration filters and renders without per-hotel special cases.
Localized texts on your terms. Marketing texts are generated per locale from templates you control and are grounded in the hotel's verified attributes, so the tone stays yours and every claim matches the facts.
Continuous quality assurance. Enriched content passes automated quality checks before it is served.
AI-Generated Content Marking
Every piece of AI-generated content we serve is marked in a machine-readable way with the same Compliance envelope, on each generated text cell and on every image. Alongside its Text, each entry in GeneratedTexts (hotel-level and per-room) carries the envelope, and each entry in Images carries the identical one. Each locale in GuestRating.Texts carries it as well.
"GeneratedTexts": {
"en-US": {
"Summary": {
"Text": "Set on a quiet stretch of the Bay of Palma, ...",
"Compliance": { "AIGenerated": true }
}
}
}
// Each entry in "Images" carries the identical "Compliance" envelope.
This satisfies Article 50(2) of the EU AI Act, which requires synthetic text to be machine-readably marked as AI-generated, so key your own traveler-facing disclosure off this flag. Texts are fully AI-written, while for images the AI-generated layer is the classification and captions we add on top of real provider photography.
How Text Generation Flows
A typical text workflow ties the three capabilities together: shape a template with saveTextTemplate, confirm the result with previewText, then kick off generateText and poll getTextGenerationStatus until it reports completion. From there, read the finished content back with getEnrichedHotel or listEnrichedHotels.
Need help? Reach our integration team at integrations@ygo.ai.
