Student-success intelligence · bounded AI

Ask your student data
anything. Trust every answer.

University Analytics turns plain-English questions into trusted student-success metrics — drawn from a governed catalog, never improvised SQL. Every number is reproducible, explained, and auditable.

Get a demo Explore the live demo

For provosts, registrars & institutional-research teams.

An institutional-research analyst asks University Analytics “which courses have the highest DFW rates?” and gets back “CHEM 101 at 31%” — the catalog drawing on Students, Courses, and Staff data.

A few things people ask

First-gen retention by facultyHighest-DFW gateway course?Grad-rate trend + next-term forecast

fall-to-fall retention◦4-year graduation rate◦dropout rate◦course DFW rate◦course fill rate◦term GPA◦probation rate◦time to completion◦student–instructor ratio◦enrollment forecast◦headcount◦first-gen outcomes◦fall-to-fall retention◦4-year graduation rate◦dropout rate◦course DFW rate◦course fill rate◦term GPA◦probation rate◦time to completion◦student–instructor ratio◦enrollment forecast◦headcount◦first-gen outcomes◦

Why teams trust it

Plain English in, governed metrics out

Ask like you'd ask a colleague. The assistant maps your question to a vetted metric from the catalog — and asks a clarifying question instead of guessing when intent is ambiguous.

It never invents SQL

Answers compile from a typed catalog through strict validation. No free-form query the model dreamed up ever touches your warehouse — so a wrong number from a hallucinated join simply can't happen.

Every answer is auditable

See the exact metric definition and the compiled query behind each result. Every question is logged — who asked, what ran, and when — so trust is verifiable, not assumed.

How it works

From a question to a number you can defend — in one step.

Ask

Type a question in plain language — “compare retention for first-gen students across faculties.”

Compile, don't improvise

The question becomes a validated spec, then a parameterized query against your conformed marts. Spec, not SQL — the model picks from a catalog, it never writes raw queries.

Trust & audit

You get the number, its plain-English definition, the exact compiled query, and a permanent audit-log entry behind it.

Trust & FERPA

Built for data you can’t afford to get wrong.

Tenant isolation by design

Each institution's data lives in its own schema; queries are scoped to it by construction — not by a filter we hope is always applied.

FERPA-minded from the start

Least-privilege access, PII redaction in logs, per-institution retention windows, and data-subject deletion built in.

Your data stays yours

Bring a read-only view shaped to a simple contract. We never improvise queries against your source systems.

Auditable end to end

Every answer carries a reproducible definition, the compiled query, and a logged trail of who asked what, when.

Isolation by schema-per-tenant · row-level access checks · audit log on every query.

See it on your numbers.

Walk through a live answer — question to compiled query to audit trail — with your own student-success questions in mind.