Destination Ideas API for Travel Sites: What It Does and Does Not Do
A traveller says what they love and what they cannot stand, and the API suggests places with reasons. It knows no flights, rooms or prices, so it belongs at the dreaming stage, before your search.
TasteRay can suggest travel destinations from what someone says, with a written reason per idea, for the inspiration stage of a travel site. It is not a booking engine: it does not see flights, rooms or prices. In this capture the model chose freely, took 27.9 seconds and checked nothing against inventory, so precompute and verify.
Travel is the one page in this section where we did not send a list. That changes what the output means. On our other pages the API ranked items we supplied, while here it proposed places from its own knowledge. Those are useful as a starting point, and you should read them as ideas until your own systems confirm them.
A week in October, from Warsaw
We described a couple in their 30s who love food markets, walking and small museums, and who hate crowds and long flights. The constraints gave a four-hour flight limit and Warsaw as origin. We sent no items, so the model chose the places.
It took 27.9 seconds, the slowest of our captures. Do not put this in front of a waiting visitor. Results came back in the order below, with the third scoring above the second, so sort by score yourself if you show it.
{
"vertical": "travel",
"context": {
"profile": "Couple in their 30s, week in October, loves food markets, walking and small museums, hates crowds and long flights.",
"preferences": [
"food",
"walkable",
"quiet"
],
"constraints": {
"max_flight_hours": 4,
"origin": "Warsaw"
}
},
"options": {"count": 3}
}meta.processing_time_ms 27,465. The model's sentences are verbatim.Ljubljana & the Slovenian Coast
95 out of 100, taste_match_score
This is the week where you stop rushing and start savoring: market breakfasts, slow river walks, and long coastal dinners with nothing louder than your own conversation. Slovenia gives you the feeling of discovering somewhere refined and intimate without sacrificing the ease you need after a short flight.
- Ljubljana is built for unhurried wandering: the compact old town, riverside cafés, bridges, and Central Market create a food-first rhythm without constant transit.
- Its small but thoughtful museum scene—including the National Gallery and Museum of Modern Art—fits your preference for cultural stops that enrich a walk rather than consume an entire day.
Bologna & Ravenna
92 out of 100, taste_match_score
You are not looking for a checklist of famous sights—you want a shared life for a week, measured in espresso, market stalls, portico walks, and dinners that become the evening's main event. Bologna lets you feel deeply immersed without being pushed through a giant-city itinerary.
- Bologna is one of Europe's most satisfying cities for food lovers: Mercato delle Erbe, the Quadrilatero, fresh pasta, mortadella, Parmigiano Reggiano, and slow trattoria meals are part of daily life.
- The city's miles of porticoes make walking comfortable even when October brings a shower, so you can stay spontaneous rather than trapped by the forecast.
Brno & South Moravian Wine Country
94 out of 100, taste_match_score
This is your antidote to crowded, overproduced travel: a quietly beautiful week where the pleasures are real and close together—market breakfasts, modernist design, vineyard walks, and wine shared at an unhurried table. It feels like choosing depth over noise.
- Zelný trh gives you the food-market ritual you love right in the center, while Brno's café culture makes lingering feel natural rather than indulgent.
- Moravian Gallery and Villa Tugendhat add genuinely memorable cultural texture in a scale that suits you better than blockbuster museums.
Fields shown: item.name, metadata.taste_match_score, explanation.why_match, explanation.key_factors.
What we left out, and what nobody checked
We removed the sentences about flight times and direct routes. We had no way to confirm them for your dates, and an airline schedule is exactly the fact a model should not state. The first reason above still mentions a short flight, so check that against your own data as well.
The response also carried budgets, durations and similar fields in its generated data. We show none of them, and you should show none either. The markets, museums and neighbourhoods named above look right to us as of this capture, but nothing in the API verified them, and you would want an editor or your own content to confirm any place claim on your site.
Two honest ways to use this
- Ideas, prepared in advance
- Run requests overnight for traveller segments and origins you care about, have an editor check the places and reasons, then publish the survivors as an inspiration page that links into your search. Latency stops mattering, and a person has read every sentence.
- Your own destinations, ranked
- Send your bookable destinations or packages in
items, up to ten, with the traveller's sentence. The API only returns items from the list you send. This is the same documented call as our other pages, and we have not captured it for travel, so test it on your own products first. - Not a live booking assistant
- For a request answering "what can I book in October?", the facts have to come from your inventory, and at this speed a visitor would wait. If you want that, the API can sit behind your search, not in front of it.
What changes in a travel integration
- Constraints people really give
context.constraintstakes any key-value limits. We sentmax_flight_hoursandorigin, and the third idea answered with a destination reachable overland. Your search must still enforce a limit, because the model only reads it.- Seasons
- Reasons here argued from October being calmer than summer. A claim about weather or crowds is the model's judgement, so keep your own seasonal data on the card.
- Rejections
- The API remembers nothing. If a traveller rejects Bologna, send it in
context.historynext time, or leave it off your list.
When to choose something else
If your need is to rank millions of hotel or route options by what past travellers booked, you want a behavioural model on your own data. Vendors such as Recombee and Amazon Personalize both mention travel on their sites. Read build vs buy before you decide.
Sources and further reading
FAQs about Destination Ideas API for Travel Sites
Can TasteRay book trips or check flight prices?
No. TasteRay has no connection to flights, rooms or prices, and it books nothing. It suggests places or ranks a list you send, with a written reason for each. Anything a traveller can act on, such as routes, dates and prices, has to come from your own inventory and booking systems.
How long does a destination request take?
Our capture on 5 November 2026 took 27.9 seconds, the slowest we saw in fast mode, where most responses took two to three seconds. It sent no list, so the model chose the places itself. Prepare travel ideas in advance and serve them from your own store, and avoid making a visitor wait.
Are the travel facts in the reasons accurate?
Treat them as unchecked. The model writes from its own knowledge, and nothing in the API verifies a market, museum or flight time against a live source. We left flight claims out of our excerpt for that reason. Have an editor review any place claim, and take routes and prices from your data.