Canon Compass

Try it in 30s: open the demo, click an example chip, then click "Find my entry point".

NOTE (STUB mode): the Qloo hackathon API key was still pending at submission time, so the live demo runs in a clearly labeled STUB mode with representative signals. Setting QLOO_API_KEY switches it to LIVE Qloo calls with no code change.

Canon Compass is a taste-aware lore agent for The Neural Wars, an original sci-fi universe (starting with Fractured Code). Big fictional worlds are hard to enter: new readers don't know where to start or why it's for them. Canon Compass asks what you already love (films, books, games, music) and finds your best entry point.

The agent resolves your tastes into Qloo entities (/search), pulls cross-domain affinities (/v2/insights), and maps those signals onto a curated, spoiler-free canon map. You get a recommended starting point, a reading order and related cultural comps, and every recommendation cites the Qloo signal behind it, so it is explainable rather than a black box. The UI shows the cited signals plus the full tool-call trace.

Qloo is the core, not a garnish: without cultural affinity data the agent cannot connect "I love Blade Runner and Radiohead" to the right door into the story. The same pattern generalizes to any franchise, publisher or game studio that wants onboarding driven by taste, not demographics, and it needs no personal data.

Challenges

The hackathon API key was still pending close to the deadline, so the app ships a clearly labeled STUB mode and switches to LIVE when QLOO_API_KEY is set, with no code change. Keeping recommendations spoiler-free while still useful meant curating a canon map of public entry points and taste clusters.

What's next

Wire the live Qloo key, add an LLM planner to the tool loop, support multi-franchise canon maps, and ship a publisher onboarding widget.

Built With

• qloo