What it does
Aftertaste plans a complete date night — dinner, film, walk-home soundtrack, and nightcap — grounded in the Qloo taste graph, from 2–4 loved artists or films plus a city. Give an agent someone's taste in music and film, and it plans their whole night in that taste language, with receipts.
How it works
- Resolves your loves to Qloo entities (GET /search).
- Plans four chained legs (GET /v2/insights), ranked by cross-domain affinity — each winner feeds its tags forward into the next leg.
- Shows its work for every pick: affinity score, driving seed, and shared taste threads.
- A "Without Qloo" counterfactual button re-plans popularity-only: without Qloo, a Radiohead fan and a Beyoncé fan get the same restaurant.
Sample-mode examples: Radiohead + In the Mood for Love in New York City → Kaiseki Den → Drive My Car → Nils Frahm → Bar Lumen; Frank Ocean + Past Lives in Chicago → Little Lantern at 0.93 affinity.
Why Qloo matters
The ranking, the cross-domain jump, and the explanations all come from Qloo affinities over 250M+ entities. The ablation proves it: without Qloo, the plan collapses to generic popularity.
How we built it
Zero-dependency Node.js with vanilla HTML/CSS/JS. The Qloo API is called with X-Api-Key, using /search and /v2/insights with signal.interests.entities and filter.type. Run with npm start; npm test runs a smoke test. The demo runs in a clearly labeled sample-data mode, and setting QLOO_API_KEY switches to the live API.
Try it
Build Disclosure: AI-built. Built by AI agents (Tree / Muse fleet) directed by Rene Turcios during the hackathon period; Rene set the direction and reviewed; agents designed, coded, and tested.
Built With
- qloo
Log in or sign up for Devpost to join the conversation.