An authenticated API for institutions integrating Cohortly cohort context into their own products. Currently in private development — the surface area, principles, and waitlist are below.
For schools, counseling departments, and platforms integrating cohort context into their own products. Authenticated, rate-limited, audited. The endpoint surface below is finalized; the live API and sandbox are scheduled to open later this cycle.
The planned v1 endpoints. Every response will carry a methodology version header so research outputs stay reproducible across cycles. None of these are live yet — waitlisted institutions get sandbox keys at private-beta.
Four constraints that will hold across every endpoint from day one. They constrain the product, not just the policy doc.
We will never serve an individual record through any access tier. Every endpoint enforces a minimum cohort size (n ≥ 30) and rejects queries that would resolve to a single applicant. This is a hard constraint of the API itself, not a policy you can request an exception to.
Cohort statistics are derived from records that were anonymized at intake. Names, addresses, schools-of-origin, and any free-text identifiers are stripped before scoring — before the data ever reaches the model that powers the API.
Every dataset and endpoint will carry a methodology version. When the model retrains or the feature schema changes, the version bumps and old responses remain reproducible for at least eighteen months. Citations should pin the version.
Institutional keys will generate per-request audit logs scoped to the customer. Default rate limits are 60 req/min per key with burst tolerance; quotas will be visible on the dashboard and configurable on the contract.
A non-exhaustive list of what waitlisted institutional partners have already pitched — useful as a sanity check for your own application.
We're admitting waitlisted institutions to the private beta in cohorts. Each one ships with a sandbox key, a DPA, and a kickoff call — in priority order.