Your city’s next hit is already open somewhere else.

Pick a city. Every number links to the source it came from.

Find taste twins
Count the gap
Try to prove it wrong

What ismissing?

Or check one business

About 1–2 minutes, at most 40 Qloo calls. See an example result ↓

How trends travel

Trends travel between cities.

Taiwanese bubble-tea chains opened in Tokyo from 2013; by 2019 “tapiru” was a buzzword of the year. Malatang reached Japan’s Gen Z by way of Seoul and K-pop. The demand was visible somewhere else first.

Taste twin (Qloo)
Could travel this way
Real gap, but fading
Past trend

Agent log

Profile
Twins
Compare
Doubt
Score

1
Profile
What do people here love?
Show request
insights · urn:tag
2
Taste twins
Where else do people love the same things?
Show request
insights · urn:entity:locality
3
Compare
What do the twins have that we don't?
Show request
insights · urn:entity:place
4
Doubt
Is the gap real, wanted and lasting?
Show request
urn:demographics · /v2/trending (not available for places)
5
Score
Six scores and a confidence, computed from the Qloo answers (plus Tokyo ward statistics, labelled). Product fit is Claude's judgment, labelled.

How it works

Planned by Claude

1
Profile the neighbourhoods
What do the people of each neighbourhood love?
Show request
/v2/insights filter.type=urn:tag signal.location.query
2
Find taste twins
Which neighbourhoods in other cities are loved by people with the same tastes?
Show request
urn:entity:locality signal.interests.entities (the home neighbourhood), then the taste tags each locality shares with it
3
Compare supply
Count places per category within 2.5 km of each neighbourhood’s centre, the same area on both sides.
Show request
urn:entity:place filter.tags filter.location filter.location.radius
4
Doubt every gap
Does the rest of the city already have it? Do the category’s fans match the people who live here? Is interest fading? With a key, does Google Maps see the same gap?
Show request
urn:demographics, urn:entity:place filter.tags, /v2/trending (not available for places); Tokyo ward statistics; Google Maps (cross-check only)
5
Score
Six scores and a confidence, computed in code from the answers above.
Show request
no extra calls

Why this exists

Bubble tea and malatang both took off in Japan years after they were proven elsewhere. Big chains pay for research to see gaps like these early; small founders can’t. Culture Arbitrage closes that information gap with Qloo’s taste data.

Who it’s for: independent founders, franchise scouts, landlords, local economic-development teams.

Checking one business

Already know what you want to open? Type it under the question (“yoga studio”, “ramen shop”, “瑜伽”) and the agent checks only that: which neighbourhoods are short of it compared with their taste twins.

The list holds 176 business types, kept because Qloo lists at least 5 of each in at least two of the three cities (counted 2026-10-25). Anything else is matched to a place category with Qloo’s tag search first, and the log shows what it was matched to.

Six scores and the confidence

Score	Source	How it’s computed
Proven elsewhere	Qloo	min(1, places in twin ÷ 15)
Bridge audience	Qloo	share of Qloo taste tags the two neighbourhoods have in common ÷ 0.5, max 1
Core audience fit	Qloo + Tokyo ward statistics	do the category’s fans (Qloo) and the people here lean to the same age bands? In Tokyo, the residents (ward statistics); elsewhere, fans of the neighbourhood’s top tastes (Qloo). Correlation, mapped to 0–1
Local undersupply	Qloo	(twin − here) ÷ twin, × 0.6 if more than 20 exist city-wide
Product fit	LLM judgment	Claude, 0–1; not counted when pending
Durability	Qloo	0.5 + 12-month change in /v2/trending, clamped 0–1; not measurable for places on live data, so not counted

Confidence = average of the measured scores × (1 − 0.05 × open doubts, at most 0.4 off). Below 55%, the agent rules the idea out: it stays on screen greyed out, with the reasons, and is not counted as an opportunity.

With or without Claude

With Claude, the same tools are offered to the model, which decides the order, skips dead ends and writes the summary. Without it, a fixed pipeline runs the steps in order. Either way, every Qloo call is capped per run and logged.

What the agent can’t know

A place missing from Qloo may be missing from the index, not from the street. The city-wide check and the thin-data flag catch some of these cases, not all.
Results are a snapshot of today’s taste data, not a historical backtest.
Boba and malatang are survivors. Tapioca shows a real gap can fade, which is why Durability is scored.
Similarity is between audiences, not streets. Rent, regulation and footfall are out of scope.

Example from a recorded run in Tokyo (all categories, 2026-10-24, live Qloo data)

2 opportunities from 24 of 40 Qloo calls

Strongest opportunity

Bagel shop: Manhattan to Shibuya-ku

New York to Tokyo, about 10,800 km apart

Shibuya-ku has 1 bagel shop while its taste twin Manhattan (similarity 0.69, 56 shared tastes) has 38 in the sample, a gap of about 38 times. Tokyo as a whole has only 15 bagel shops, so the category is scarce citywide, not just locally.

Written by Claude

61%

Promising

How 61% is worked out

Average of 5 of 6 scores (71.4) × doubt penalty 0.85 (3 doubts) ≈ 61%. The line marks 55%: below it an idea is ruled out.

Evidence chain

Shibuya-ku and Manhattan share these Qloo tastes:
Tourist DestinationTrendyArt Enthusiasts
Q7
35%
of taste tags shared (56 of 162), #1 in New York
Q7
−37+
bagel shop places missing in Shibuya-ku
Manhattan, New York38+
Shibuya-ku, Tokyo1 (37+ missing)
All of Tokyo15 places
What the dots mean
Each dot is one place Qloo returned with this tag. Hollow dots are the empty storefronts: places Manhattan has that Shibuya-ku doesn’t.
Q12Q13
15
across all of Tokyo
Q14
31%
age match, fans vs residents
residents of Shibuya ward (Tokyo resident register, 2026-01-01)
Q17
—
trend not measurable for placesQloo's trend data covers artists, brands, films and TV, not places
60
LLM judgment product fit: “Bagels are photogenic, easy to customise with toppings, and cheap to open as a small counter-service shop that suits solo visitors.”

Numbers link to their source on the right; other sources are named.

The agent tried to prove itself wrong.

How the score moved, check by check

Gap found: at least 38 in Manhattan, 1 in Shibuya-kuQ12Q1389%
Checked the rest of Tokyo: 15 placesQ14−584%
Checked who it is for: 31% age matchQ17−1371%
Final score, with product fit 60%, 2 doubts of its own−1061%

Doubts still standing (3)

?Bagel fans skew female and 55+ (positive 55+ and 45-54 affinity), while Shibuya-ku residents lean toward 35-44 and 45-54, so the core audience is only a partial match.
?The New York count is from a capped sample, so the true twin count may differ.
?15 exist elsewhere in Tokyo, so the gap may be local only.

Together they multiply the average score by 0.85.

Six scores

Proven elsewhere100
Bridge audience69
Core audience fit31
Local undersupply97
Product fit60
Durability—

What each score means

Proven elsewhere Qloo
at least 38 places in Manhattan. 15 or more places = 100.
Q13
Bridge audience Qloo
35% of Qloo taste tags shared by Shibuya-ku and Manhattan (half shared counts as a full match). Measured as the shared-taste score: the share of Qloo taste tags both neighbourhoods have, ÷ 0.5, max 1. Cross-interest bridges (like K-pop → malatang) are not measured yet.
Q7
Core audience fit Qloo
Age profile of the category's fans vs residents of Shibuya ward (Tokyo resident register, 2026-01-01). Age overlap only; gender is not used yet.
Q17
Local undersupply Qloo
1 in Shibuya-ku vs at least 38 in Manhattan, 15 across Tokyo. Reduced when the whole city already has more than 20.
Q12Q13Q14
Product fit LLM judgment
Bagels are photogenic, easy to customise with toppings, and cheap to open as a small counter-service shop that suits solo visitors. Claude’s judgment: photogenic, customisable, health image, solo-friendly, cheap to open. Not Qloo data. Included in confidence.
Durability Pending
Not measurable: Qloo's trend data covers artists, brands, films and TV, not places. Bubble tea in Japan is the warning case: a real gap that faded. Not counted in confidence.
not fetched

GitHub repository README MIT License