TasteFit
Open the map

Who should rent this unit?

Pick a storefront. An agent reads what the street already loves on Qloo's taste graph, checks what is already served nearby on open map data, and the rent each concept can carry.

Pick a unit on the map

Built on Qloo's taste graph, OpenStreetMap and official business registers. Open source under MIT. The plan and comparison on this page are sample data.

TasteFit is a hackathon prototype. Follow the project on GitLab.

Sample output. Street, places and scores are invented.

How it works

  1. Pick the unit

    Choose an example point or drop a pin on any storefront. Add the asking rent if you have it.

  2. Watch the agent ask Qloo

    Each lookup appears as it happens: what it asked, what came back, which tag it chose. About 20 calls, a few minutes.

  3. Take the memo to the meeting

    A one-page shortlist with a GO, ADJUST or NO-GO call per concept, the evidence behind it and the limits. Prints cleanly.

Same model, same question. One of them can check.

Every run asks the language model alone first, then marks which of its picks the evidence confirmed, contradicted or never tested.

Without Qloo

  • Specialty coffeeContradictedMore cafés nearby than the city's mix predicts.
  • Brunch caféNot testedNo taste evidence either way.
  • Neapolitan pizzaNot testedNo taste evidence either way.
  • Korean fried chickenConfirmedQloo ranks it 1 of 6 here.

With Qloo

  1. Korean fried chickenGORank 1 of 6, 2 nearby, overlap 0.64.
  2. Lebanese grillGOMissed by the model alone. Rank 2 of 6.
  3. Ramen shopADJUSTLoved, but 7 places nearby. Negotiate rent.
  4. Mediterranean bowlsNO-GO9 places nearby, interest fading.

Sample comparison. In the live app both columns come from the same run.

What it does not know

Qloo describes the aggregate taste of an area's audience, not foot traffic, sales or any person. Scores rank the candidates in one run against each other. Nearby supply counts OpenStreetMap venues, which can be incomplete. Rent ceilings come from a survival model built on official data, calibrated for Paris and Chicago only. Alcohol-led concepts are left out.

City Scan and rent ceiling

Prebuilt from open map data, separate from the taste check. A model, not a forecast.