Reel
Your taste, played back.
- Type
- AI agent + taste data
- Year
- 2026
- Role
- Product idea, agent design, full-stack, visual design, launch video
- Stack
- Next.js 16, React 19, TypeScript, Qloo API, LLM tool calling, Vercel
Name up to five things you love. Reel asks the Qloo taste graph what else you would love, names your taste like a film title and plays it back as a reel of cards. Built for the Qloo Agentic Hackathon.
Ask a chatbot what else you would like if you love Hindia and Spirited Away, and you get the same safe list it gives everyone. Ask the Qloo taste graph and you get Reality Club, Danilla and Rumahsakit, each with a score, because it knows what people who love those two things actually love.
Reel makes that difference something you can feel in a few seconds. Your favourites go to Qloo once for each of six domains: music, film, TV, books, podcasts and brands. The reel plays back as full-screen cards: a title card that names your taste, one card per domain, the themes that run through them, rolling end credits and a story-sized card to save.
Then you can ask it anything, like a film for a rainy Sunday. An agent turns the mood into Qloo tags, asks for matches from your favourites, checks whether they fit and tries again if they don't. It can only answer with what the taste graph returned, and every step it takes is shown as it happens.
Details worth a look
- A landing page whose hero is a live wall of posters from Qloo, recast when you type a favourite
- Six Qloo calls per reel, run three at a time, with a film-leader countdown that shows each real call
- The thread: themes and moods that cross the most cards, from Qloo's own descriptive tags
- An agent with four tools that picks its own searches and cannot invent a title
- A 1080 by 1920 story card drawn in the browser, ready to post
- Burgundy and bubblegum like cinema velvet, Bebas Neue against EB Garamond
Next project
Pulen