Developers
Course generation, as an API.
The same research-grounded, guardrailed generation pipeline that powers StudAI Loop — behind scoped keys with spend caps you control.
Endpoints
The public API surface, exactly as implemented. Long-running work is asynchronous by contract.
POST /api/public/v1/courses → 202
Start asynchronous course generation for a topic. Returns a job ID — generation is long-running by design and never blocks your request thread.
GET /api/public/v1/courses/{jobId}
Poll job status and retrieve the generated course structure when ready.
POST /api/public/v1/concepts/explain
Generate a grounded explanation for a concept, using the same guardrailed generation pipeline as the product.
GET /api/public/v1/graph/concepts
Query knowledge-graph concepts and prerequisite structure for a domain.
Platform guarantees
Scoped API keys
Keys are created in the developer workspace by organization admins, carry explicit scopes, and are shown exactly once at creation.
Spend caps per key
Every key has daily and monthly AI-spend caps, visible in the usage dashboard. A leaked key cannot produce unbounded cost.
Private graph namespaces
Optionally isolate the concepts your integration creates into a tenant-private graph namespace, invisible to other tenants.
Honest error semantics
Machine-readable error codes, correlation IDs, and 429s with budget context. Rate limits are enforced per key and per plan.
Getting access
API keys are managed in the developer workspace by Institute organization admins.
Frequently asked questions
- How do I authenticate?
- Bearer authentication with an API key created in the developer workspace (organization admins on the Institute tier). Keys carry scopes and spend caps.
- Are there SDKs or webhooks?
- Not yet. The API is plain HTTPS + JSON with async job polling for long-running generation. SDKs and job-completion webhooks are roadmap items and are not sold as current features.
- Is there an uptime SLA?
- No contractual SLA is offered today. Platform health is reported honestly at the readiness endpoint, and spend caps plus rate limits protect both sides.