Inspiration
Businesses often know the demographic they want to reach, but not the culture that actually moves that audience.
A generic AI model can produce plausible recommendations, but plausibility is not the same as cultural fit. Taste was built to close that gap by grounding business strategy in Qloo's real taste graph.
What it does
Taste turns a business brief into a Qloo-grounded cultural strategy.
A user enters:
business type location target audience objective a few known cultural signals
The agent then resolves those signals to Qloo entities, explores related taste across multiple categories, identifies cross-domain cultural intersections, and translates the evidence into:
Cultural DNA campaign direction partnership ideas content strategy experience concepts explainable Qloo evidence
The current product explores music, local places, brands, film, and travel.
How Qloo powers the product
Qloo is the cultural intelligence layer, not a decorative API call.
Taste uses Qloo to:
resolve human-readable interests to real Qloo entities analyze the combined seed profile query cross-domain recommendations through Qloo Insights request explainability for those recommendations identify useful co-signals across categories
The core principle is:
Qloo maps the taste. Taste finds the overlap.
Without Qloo, the system would have to guess cultural relationships from a prompt. With Qloo, the strategy is grounded in structured taste data.
Agentic workflow
Business brief → domain planning → Qloo entity resolution → Qloo taste analysis → cross-domain Insights → explainability parsing → cultural intersections → business strategy
The agent dynamically changes which cultural domains it prioritizes based on the business context.
For example, hospitality emphasizes local places, travel, music, brands, and film. A fashion brief shifts toward brands, music, film, and creator-adjacent culture.
Cultural Intersections
The key differentiator is not just recommendation.
Taste looks for where independent Qloo signals become more useful together.
Instead of only saying:
"Your audience likes A24."
the system can reveal a film × music, brand × travel, or local-place × music intersection and convert that combination into a campaign, partnership, or real-world experience.
These are presented as co-signals, not causal claims.
Explainability and integrity
The product preserves the Qloo evidence trail.
Users can open "Why this signal" to see:
the returned entity affinity seed context available explainability data
The system does not fabricate missing evidence.
If a cultural seed cannot be resolved, it remains unresolved. If a domain returns no data, the system does not invent replacement recommendations.
Design
The product is designed as a premium cultural-intelligence interface rather than a chatbot.
The live experience includes:
a guided brief composer three demo presets a strategic result hero Cultural DNA a cross-domain affinity landscape featured cultural intersections strategy translation an inspectable agent trace
Demo presets
The live app includes three distinct briefs:
Casa Forma Boutique hospitality in Guadalajara
NOIR//FORM Independent fashion in Los Angeles
LUNA SOCIAL Restaurant and nightlife in Austin
These demonstrate that the agent changes its exploration and strategy based on the business context.
Tech stack
Next.js React TypeScript Zod Qloo API Vercel GitHub
What we learned
The biggest lesson was that an AI strategy system becomes far more credible when the model is not responsible for inventing the underlying cultural relationship.
Qloo provides the affinity. The agent provides the interpretation.
That separation made it possible to build a system that is creative without hiding where its recommendations came from.
What's next
Future versions could add:
richer audience segmentation creator discovery geographic comparison saved cultural profiles campaign testing collaborative agency workspaces additional Qloo domains and entity types
The larger vision is a cultural operating system for businesses that need to understand not just who their audience is, but what that audience is connected to.
Built With
- github
- next.js
- qloo-api
- react
- typescript
- vercel
- zod