Know what your visitors love before they walk in.
For cafés, homestays and tours in Vietnam. Pick the visitors you want — people from Seoul, Tokyo, Sydney — and get a one-page plan built from Qloo's taste graph: where they already go in your city, what those places have in common, music both sides know, and how your own ideas measure up. Every claim links to the Qloo result behind it.
Agent at work
Do this next
What their favourites have in common
Your ideas, measured
Music both sides know
Same question, no Qloo
We asked the same model the same question with no data, then looked each of its picks up in Qloo.
Places it suggested
Artists it suggested
Its advice
What this does not tell you
Evidence ledger — every cited fact and the Qloo request behind it
Agent trace
Build a brief for your place
Or open a finished example:
How it works
- Qloo first. The server asks Qloo for places in your city ranked by the taste of people living in the visitor city, and the same lists without that signal. The difference is what that market adds.
- What sets their favourites apart. Tags (ambience, setting, offerings) clearly more common among the market's favourites than among the city's top places.
- An agent maps your ideas. NVIDIA Nemotron calls Qloo tools to find the tag for each idea, then scores it against the same lists — places, music or TV. It can only use tag ids Qloo returned.
- A plan that cites its evidence. The model writes from a fact sheet; actions without a valid citation, actions that push an idea the data does not support, and names that are not in the data are removed. Names and numbers on the page come from Qloo results.
- Also an MCP server. The same tools are served at
/mcpfor any agent to call.
Qloo results describe the aggregate taste of people in a city. They say nothing about any individual guest, and no personal data is sent to Qloo.