Inspiration

Moving to a new city is usually treated as a logistics problem: housing, transport, paperwork. But the harder question is often emotional—how do I find the places, sounds, stories, and small rituals that make an unfamiliar city feel like mine?

Search engines optimize for popularity. Home2Here starts with personal taste. I built it to help newcomers translate the culture they already love into a practical path toward belonging.

What it does

A user selects their origin and destination, adds cultural anchors such as artists, films, books, brands, restaurants, or places, and chooses a budget and discovery pace. Home2Here then:

• Resolves free-text favorites into verified Qloo entities
• Builds a shared Taste DNA from those anchors
• Uses Qloo Insights to translate that signal across places, music, and stories
• Applies destination, eligible price, and first-week suitability constraints
• Explains why each recommendation connects to the user's existing taste
• Converts the results into a manageable first-week belonging plan
• Replans from “More like this” or “Not my vibe” feedback
• Shows an inspectable agent trace and clear source boundaries

How I built it

Home2Here uses browser-native JavaScript and CSS with a dependency-free Node.js HTTP server. The server keeps the Qloo credential private, resolves category-aware entities through Qloo search, and runs parallel Qloo Insights queries for places, artists, and films.

The agent filters candidates by destination and budget, turns cultural affinity into bridge explanations, and sequences the strongest results according to the selected discovery pace. A small cache and request budget protect the hackathon API allowance. The application is deployed on Render and requires no account, paid model, database, or tracking service.

Why Qloo is essential

Qloo is not a decorative recommendation widget here—it is the cultural reasoning layer. A general model can write a pleasant relocation list, but it cannot reproduce Qloo's cross-domain affinity graph from a person's real cultural anchors. Home2Here separates Qloo-supplied entity resolution, ranking, and explainability from the constraints and planning added by the application.

Challenges

The hardest challenge was making recommendations feel personal without overstating the evidence. Names can be ambiguous, Qloo responses vary by entity type, and a culturally relevant place may still be impractical for a first week. I addressed this with category hints, conservative entity matching, destination and price validation, explicit confidence labels, and source-aware explanations. Another challenge was building a meaningful second turn: feedback changes the Qloo signal or exclusions and forces a fresh plan rather than merely changing the interface.

Accomplishments

• All nine anchors across three judge-ready relocation personas resolve against the live Qloo catalog
• Every persona returns all three cultural domains and a four-step first-week plan
• Live and illustrative modes are visibly distinguished
• The Qloo API key never reaches the browser
• Automated tests cover resolution, filtering, orchestration, validation, fallback disclosure, static delivery, and the complete plan endpoint

What I learned

I learned that strong cultural recommendations require both affinity and boundaries. Taste intelligence identifies meaningful connections; a useful agent must also preserve uncertainty, respect practical constraints, explain its reasoning, and turn discovery into action.

What's next

Next, I would add collaborative household profiles, accessibility and dietary constraints, neighborhood-level transit context, saved journeys, and longitudinal feedback so Home2Here can learn which recommendations actually became part of someone's new life.

Built With

• agentic-ai
• css3
• cultural-intelligence
• html5
• javascript
• node.js
• qloo
• qloo-api
• recommendation-engine
• render
• taste-ai