Live product
Gravity English
Tenant-aware learning infrastructure with a durable AI handoff.
Product boundary
The problem being solved
Schools, teachers, and enterprise customers need distinct branding, membership, curriculum, and domains without requiring a separate application build for every tenant. Long-running curriculum generation must also remain outside the interactive learner request.
Verified architecture
System decisions visible from public-safe evidence
- SvelteKit
- Svelte 5
- TypeScript
- Node.js
- PostgreSQL
- Drizzle ORM
- Clerk
- AWS S3
-
Request-time host resolution maps root domains, tenant subdomains, and verified custom domains to organization, branding, localization, membership, role, and subscription context.
-
Clerk supplies authentication while PostgreSQL and Drizzle retain application-owned user, organization, membership, role, and job authority.
-
Curriculum generation is committed as organization-scoped work for the separate Gravity Swarm Architect service instead of blocking an interactive learner request.
-
Larger blueprint and media payloads use AWS S3 while relational records preserve tenant ownership and orchestration state.
Current evidence
What the implementation establishes
-
One application resolves tenant-specific brand, localization, organization, role, and subscription context for each request.
-
Persisted generation jobs keep long-running curriculum and thumbnail work outside the learner-facing request path.
-
Typed server-defined curriculum can evolve independently from compiled interface deployments.