Product tour
Everything an institution needs to run quality assurance, not just collect surveys
OneRubric is deliberately narrow: not an LMS, not a student-information system, not a research-analytics platform. It is the system that takes an evaluation from submission to a verified improvement.
7
Role tiers
Student, lecturer, head, dean, executive, QA directorate, platform operator
≈250
Purpose-built screens
Each scoped to one role’s data
±0.20
Verification threshold
Improved, regressed or no change, stamped automatically next period
By role
Pick a seat at the table
Access is enforced by scope rows binding a user to specific schools and programmes. A hand-typed URL to a course outside your scope returns a 403, not a partial page.
A guided survey, and proof that feedback changed something
Students rate each course on a five-point scale across six areas of teaching, add anonymous comments, and can later see what their lecturer did in response.
- Multi-step evaluation wizard with progress saved automatically
- Anonymity reminders at every step; one evaluation per course
- Course expectations set at the start of term and answered at the end
- A “you said, we did” feed of concrete changes lecturers made
- Personal evaluation history and in-app notifications
Private AI
Reads every comment. Cites its sources.
Four AI surfaces, all built on aggregate-only digests and an anonymised comment array. No student identifier reaches the model, whichever provider runs it.
- Chat with retrieval. Ask what students say about pacing; the assistant retrieves the matching anonymous comments by meaning and answers with a clickable Sources strip and similarity scores.
- On-demand briefings per scope: institution, programme, faculty or course, streamed as they are written.
- Narrative reports and period briefings generated automatically when a period closes, replacing the hand-written cycle.
- Outline verifier: checks an uploaded course outline against the institution’s required structure and returns pass or fail with a rationale.
- AI flag briefs summarise flagged integrity items to speed up human triage; they never act on their own.
Self-hosted: Mistral 7B via Ollama on your hardware. Hosted: Together.ai by default, OpenAI optional, each under a no-training DPA. Swappable from the operator UI; metered monthly; degrades to “AI unavailable” rather than erroring.
The action loop
A commitment, a baseline, and a verdict next period
The feature that turns evaluation from a compliance chore into a measurable improvement programme.
- 1
Flag
An aspect is flagged for a course: an outlier, a decline, or an AI red flag.
- 2
Commit
The lecturer writes concrete action items and submits the plan.
- 3
Baseline
The aspect’s current average is recorded, keyed to lecturer and course so it survives roster changes.
- 4
Approve
The head of department reviews: draft → submitted → approved → completed.
- 5
Verify
A scheduled job re-measures the same aspect in a later period: +0.20 improved, −0.20 regressed, otherwise no change.
- 6
Report
Leadership sees the closure rate: how many flagged aspects have a plan, how many verifiably improved.
Early warning
Signals before the term is over
End-of-term evaluation becomes confirmation, not surprise.
Mid-course pulse
A short formative check-in mid-semester, private to the lecturer and their head or dean. Pulse results never feed the leaderboard.
Course expectations
Students say what they hope for at term start; lecturers record met, partially met or unmet at term end; students see the response.
Outline compliance
Which lecturers have uploaded a compliant course outline, AI-verified against the required structure, tracked per period.
Anomaly detection
Courses whose rating dropped sharply relative to their cohort, with a small-cohort fallback, so an institution-wide hard semester is a trend and not a wall of alarms.
Reporting and exports
One click to a board-ready PDF
Every document carries the institution’s branding from the same theme source as the live app, so exports never drift from the screen.
- Executive board pack: headline versus targets, biggest movers, programme accountability, an equity lens, exemplars, action-loop closure and a student-voice section, with a credibility gate when the sample is thin.
- Per-course and comments-only PDFs; comments from upheld disputes are excluded automatically.
- Bulk period export: a background job packages a whole period (summary CSVs, roster, a PDF per course) into a ZIP and emails the link.
- Audit-log CSV and a full tenant JSON export on demand.
Runs itself
The system tells you when to look
A bell feed is the source of truth; email is best-effort with per-user opt-out and a logged outbox with retry. Eleven scheduled jobs keep the institution’s tempo, each with a heartbeat.
Scheduled automation
- Auto-close elapsed periods
- Period-close alerts to the QA office
- Reminders to students and faculty
- Weekly digest
- Nudge faculty to close open expectations
- Verify action plans against new data
- Send period briefings
- Process bulk exports
- Backfill AI embeddings and flag briefs
- Purge expired trials
Identity and integrations
- Google Workspace and Microsoft Entra sign-in
- SAML 2.0, SP- and IdP-initiated, with replay protection
- SCIM 2.0 user and group provisioning from Okta or Entra
- Magic-link invitations
- Idempotent Canvas and Moodle sync of terms, courses, sections, enrolments
- Roster and people import from spreadsheets
White-label from admin screens
- One primary colour derives the whole ramp; no rebuild
- Logo, crest, product name, hero treatment, density
- Schools, programmes, campuses, period and role labels
- Custom evaluation questions per programme
- Timezone, locale, email domains, AI persona
- A guided setup wizard brings a new install online
SAML + SCIM
Enterprise identity
Canvas · Moodle
LMS sync
5
Cloud deploy targets
DigitalOcean, AWS, GCP, Azure, Railway; or Docker Compose for evaluation
Ready to see it on a sample institution?
A detailed feature catalogue is available for procurement on request.