Deadline week hits the inbox first: Celeste for admissions and registrar

Every admissions office knows the calendar shape: the inbox is manageable in October, and then a deadline approaches and four hundred versions of the same six questions arrive in a week. What are the requirements for this program. Is the deadline postmarked or received. Do my community college credits transfer. When does registration open. What is the FAFSA code. The staff answering is the same size in deadline week as in July, and applicants are anxious in inverse proportion to how fast answers come back.

Surge-shaped, repetitive, and already answered in your published materials: that is the profile Celeste is built for.

Published answers, delivered when asked

Ingest what the institution already publishes: admissions requirements by program, the academic calendar, the transfer credit guide, registration procedures, aid application basics. The agent answers on email and website chat from what those materials actually say, at midnight before the deadline, the four hundredth time exactly as patiently as the first.

The transfer credit guide is a favorite example because it is genuinely useful and nobody can find it. “Does my statistics course transfer” answers from the published equivalency list when it is on the list, and when it is not, the agent says the published guide does not cover it and routes the question to the evaluators, logging a named knowledge gap. Ask-frequency turns into a ranked list of what your published materials are missing, which is a report no survey will give you.

Policy is automatable. Determinations are not.

The line this desk never crosses: it answers what the published policy says, and it never predicts what the institution will decide. “What is the aid deadline” is a policy question. “How much aid will I get” is a determination, and it escalates to a human, always, with the context gathered. Same split for admissions chances, appeal outcomes, and residency classifications. The boundary is enforced by topic, not by tone: determinations are outside every topic you allow, and messages that cross into them are held for staff.

Certain words should never meet an automated reply on a campus. “Appeal,” “withdraw,” “emergency,” and anything your student affairs office designates go on the escalation keyword list, and a message containing one is barred from an autonomous reply outright, held with the trigger flagged. And this desk answers questions about published policy rather than looking into individual student records, which keeps it useful without wandering into record-privacy territory. We make no compliance claims, and the design keeps the desk on the public side of that line.

Piloted one office at a time

An institution does not turn this on campus-wide. Start with one queue, admissions is the natural pick, in shadow mode, where the agent drafts silently against real inquiries while staff read what it would have said. Requirements and deadline questions graduate first. Transfer credit answers go live once the equivalency material has proven complete enough. Anything the registrar wants held stays held; every topic has its own dial.

What the institution buys is deadline-week capacity without deadline-week hiring, and something less measurable: the applicant who got their answer at midnight applies, while the one who waited four days sometimes does not. The staff, meanwhile, spend deadline week on the cases that need a human, which were always in the pile, buried under the six questions.

See how it works or the product. The trial runs silently against a real inbox first, so the office can grade a deadline cycle of drafts before anything sends.