Browse games by category

Categories and the assignment pipeline

The catalog grew past the point where a single swipeable feed could surface the right game, so we added a real category system: Puzzle, Arcade, Shooter, Platformer, Sports, Word, and the rest, each backed by its own data model and endpoints rather than a loose tag column.

The core of this release is an agent-driven assignment pipeline. We shipped category endpoints plus an icon-generation endpoint, and wired category assignment into the game launch path. When a game goes live, the agent classifies it into the appropriate category automatically and produces a category icon, so the shelves stay accurate without creators filling out metadata forms or operators triaging a tag backlog. This is the same worldview that drives the studio: let AI absorb the boilerplate work so creators only have to make the game.

We deliberately chose deterministic categories over freeform tags. Tag soup decays the moment nobody maintains it, and a stale taxonomy makes discovery worse, not better. Assigning at launch keeps classification consistent and queryable: a category page is a real query against assigned games rather than a fuzzy text match.

Category SEO and landing pages

Each category is a standalone landing page rather than a filter that only exists inside the app shell. These pages are built to be indexable on their own, with category-specific metadata and titles, so a search for a genre can land directly on the relevant Remix shelf. Standalone category routes give search engines real URLs to crawl instead of one opaque feed, which is how organic discovery actually reaches new players.

Search also got a proper home in browse: a query can resolve straight to a game or jump to a genre page rather than forcing a player to swipe until something matches. The endpoints powering category pages and search share the same assigned-category data, so what a player browses and what a crawler indexes stay in agreement.