Everything around the examination.

Eligibility is modelled, not a free-text form. You define requirements of five kinds — a credential or license, an education qualification, supervised practice hours against a minimum, a training or continuing-education completion, or a declaration the candidate signs — each with its own label, detail and order. Requirements are retired rather than deleted, so a decision made under an old rule stays explicable years later.

A candidate applies and their application stores the requirements as they were on the day they applied. Changing your rules next year cannot retroactively invalidate a pending application. Answers are structured per requirement, documents attach per requirement rather than as one undifferentiated bundle, and your reviewers can request a specific missing document against a specific requirement, with the request tracked to fulfilment.

Approval is time-boxed: a decision carries an approved-until date derived from your program's eligibility validity period. And the whole thing is optional — a program set to open lets anyone register and never shows a reviewer a queue. For a program that asks more of a candidate than a checklist, Application Studio designs the form, the journey and the review.

An examination window is a dated period with a delivery mode, an optional seat capacity and an optional registration deadline. Inside it you publish appointments: a start instant, a duration, a capacity and a place. Times are stored as UTC instants and displayed in a named zone, so an appointment at 09:00 Eastern renders correctly on both sides of a daylight-saving boundary — storing an offset instead is right until March and wrong afterwards, and the cost of that error is a candidate arriving an hour late for an examination they cannot retake for ninety days.

Capacity is enforced by the database on both the way a seat is taken and the way it is moved, because a rule enforced only on insert is not a rule — rescheduling moves a seat by update. What it costs a candidate to change their mind is a policy you set, tiered by how much notice they gave, and the amount is frozen at the moment the change is decided so that editing the policy next month cannot restate what somebody was charged.

A program that tests continuously — any business day at a center, or on demand online — schedules appointments directly, without a year-long window with invented dates and a made-up deadline that a candidate would read as a commitment.

A candidate asks for a testing accommodation and your body decides it, with a reason. What a proctor, a roster, a staffing board and the delivery clock learn is the operational fact — this candidate has time and a half, a room of their own, a reader. None of them learn why, because none of them need to, and every copy of a reason is a copy that can leak. The candidate's statement and the decision are readable by the candidate who wrote it and by the people at your body who may decide, and by nobody else, including the examiner who will run the sitting.

The decision is applied where it matters: the delivery clock builds the sitting's deadline from it, and an appointment too short for it is refused rather than booked and discovered on the day.

A body's own systems read its records through named keys with the scopes they need — credentials and their holders, registrations and appointments, released results — over a read-only API that returns identifiers, statuses, dates, outcomes and the candidate's public id, email and name your body already holds, and never an answer, a question or a payment detail. A key is shown once, may expire, records when it was last used and is revoked with a reason.

Outgoing webhooks tell your systems what happened, as it happens: each delivery is signed with a secret you hold, retried with backoff, paused after repeated failure with the reason, and replayable by hand. Endpoints must be HTTPS with no credentials in the address; a test event is a click away, and the last deliveries sit beside their status codes.

A body running a program gets the answer it never had: of the people who looked, how many applied, registered, sat, passed, were certified and renewed, and where they fell away. Every stage after the first is already a row the platform writes as the journey happens; the first — a visitor reading your page before signing in — is counted with no person, no address and no query string attached. The operations dashboards do the same for the staff who run the platform: pipeline, queues, quality, and what needs attention today.