Evaluations that end in evidence.

Students rate and comment. Lecturers commit to a plan. Next term OneRubric checks whether the score moved, and your council sees the result.

Designed with the quality-assurance directorate of a public university in Ghana.

The whole evaluation cycle. In one place.

Take the product tour

Built around what happens after the survey.

of comments read and grouped: 100%
roles, each with its own view: 7
areas of teaching rated: 6
the rule behind every verdict: ±0.2

The QA office’s view of the whole institution.

Coverage, ratings, flagged aspects and what last term’s plans achieved, on one screen.

EUExample UniversityQuality Assurance Office

Overview

Evaluations

Comments

Action plans

Reports

Settings

Signed in as Director, QA

First semester 2026/27 · closes in 9 days

Overview

All schoolsExport board pack

Response coverage

71%

Target 70% · met

Average rating

4.02

+0.11 on last term

Flagged aspects

20 with an action plan

Verified improved

60%

12 of 20 plans due

Action loopLast term’s plans, measured again

12 of 20 action plans improved the score they targeted.

Improved 12No change 5Regressed 3

Flagged aspects with a plan20 / 24

Worth a lookBelow their cohort

ACC 302Cost AccountingAssessment feedback · no plan yet−0.6

PAD 105Public Administration IPacing · plan in review−0.4

ECO 201Intermediate MicroeconomicsWorkload · plan approved−0.3

Recent verdicts±0.2 on the five-point scale

Course	Aspect	Baseline	This term	Verdict
MGT 214Organisational Behaviour	Delivery clarity	3.6	4.3	Improved +0.7
ECO 104Principles of Economics	Feedback turnaround	3.2	3.5	Improved +0.3
HRM 310Employee Relations	Workload	3.9	3.8	No change −0.1
STA 102Business Statistics	Pacing	3.7	3.4	Regressed −0.3

Overview

Shown with example data from a demo institution. Heads, deans and the executive each see the same view for their own scope.

Evaluation used to be a filing exercise. Now it’s evidence people trust.

Small classes protected

Scores and comments stay hidden until enough students respond, so no one in a small class can be singled out.

Fair to lecturers

A lecturer can dispute a comment that breaks the rules. The head of department decides, and an upheld comment disappears everywhere.

AI where you choose

On your own servers with no outside AI traffic, or hosted under a no-training, zero-retention agreement. It never sees a student’s name.

Every decision on the record

Each privileged action is written to an append-only audit log that exports for procurement or accreditation.

Read the security overview

OneRubric was designed alongside the quality-assurance directorate of a public university in Ghana, starting from the way its evaluation cycle actually ran.

Now open to the first cohort of institutions.

Read about the partnership

Live in a day. Measurably better by next term.

Three ways to run it. The same software.

Pricing will be announced at launch. Every option includes a full export of your data at any time, free.

Compare the three models

Questions QA offices ask first.

Can lecturers or administrators see who wrote a comment?

No. A review row holds the comment and rating only; it carries no student email. Course aggregates and comments are withheld until a configurable number of distinct responses is reached, so a small class cannot be reverse-engineered. The AI receives aggregates and an anonymised array of comments, never the identity column.

Where does the AI run, and does it learn from our data?

Self-hosted deployments run Mistral 7B through Ollama on a server you control, with no outbound AI traffic. On the hosted service, inference runs at Together.ai by default (OpenAI optional, off by default) under a DPA with zero-day prompt retention and a contractual prohibition on training. In both cases only anonymised aggregates and comments are sent.

What does “verified next term” actually mean?

When an action plan is created it records the flagged aspect’s score as a baseline. A scheduled job later compares the same aspect, for the same lecturer and course, in a later period: a rise of 0.20 or more is stamped improved, a fall of 0.20 or more regressed, anything else no change. Leadership sees the aggregate closure rate.

How do people sign in?

Through your institution’s identity provider only: Google Workspace, Microsoft Entra (restricted to your directory), or SAML 2.0 for Okta, Entra, KeyCloak, ADFS and Shibboleth. SCIM 2.0 automates provisioning and de-provisioning. There is no email-and-password sign-in.

Can we self-host instead of using the hosted service?

Yes. The same software deploys to DigitalOcean, AWS, GCP, Azure or your own on-premise infrastructure. In that model OneRubric the company has no sub-processors and no operational access to your data without an explicit, time-bounded support grant.

What happens to our data if we leave?

You can export a complete, portable copy of your institution’s data at any time, free, regardless of contract status. On the hosted service, tenant data is retained for 90 days after termination so you can request a final export, then purged. Self-hosted customers already hold their own database.

Do you hold SOC 2 or ISO 27001?

Not yet. We follow the practices but do not hold either audit report; both are on the roadmap. Every contract includes source-code access and the right to commission an independent security audit each year. We would rather say this plainly than imply a certification we do not have.

How long does it take to get started?

On the hosted service you sign up, walk a short setup wizard (institution name, sign-in providers, branding) and invite your QA team the same day. A self-hosted production deployment takes roughly fifteen minutes of setup with the provided orchestrator, after which the same wizard brands it and brings it live.

What does it cost?

Pricing will be announced at launch. If you are evaluating now, tell us your student count and deployment preference on the contact page and we will walk you through what each model involves.

Course evaluation, rebuilt around what happens next.

A 30-minute walkthrough on a sample university, or a 7-day trial on the hosted service with no card required.

Book a demo Start a trial

Course

Aspect

Baseline

This term

Verdict

MGT 214Organisational Behaviour

Delivery clarity

3.6

4.3

Improved +0.7

ECO 104Principles of Economics

Feedback turnaround

3.2

3.5

Improved +0.3

HRM 310Employee Relations

Workload

3.9

3.8

No change −0.1

STA 102Business Statistics

Pacing

3.7

3.4

Regressed −0.3