
Free SPECS professional workbook
AI-Assisted Grading Governance Test
Test whether an institution has retained meaningful human accountability before AI influences marks, feedback or progression-related decisions.
About the tool
A practical working resource for evidence-led improvement.
A structured governance and assurance workbook for universities considering or reviewing AI-supported marking, grading, moderation or feedback. It tests academic authority, assessment validity, context-specific validation, fairness, data protection, meaningful human review, student rights, monitoring and change control before the institution reaches a deployment decision.
Inside the workbook
What the tool provides
The workbook combines structured questions, traceable evidence and accountable follow-through in one editable working format.
- 01Twelve grading-governance criteria
- 02Five maturity levels
- 03Five decision signals
- 04Evidence and reviewer-rationale fields
- 05Non-negotiable stop conditions
- 06Bounded-pilot rules
- 07Owned improvement and closure fields
- 08Automatic institutional disposition
Fictional worked example
See how the workbook works
The images below use fictional institutional information to demonstrate the workbook’s structure, evidence trail and management summary.

The summary keeps stop conditions, limited-pilot restrictions and unresolved actions visible to decision-makers.

Each judgement is linked to expected evidence, a reviewer rationale, an accountable action and closure evidence.
Illustrative use only. The examples do not represent a real institution, accreditation judgement or regulatory decision. Each workbook must be aligned to the applicable framework and reviewed using qualified human judgement.
Before you download
Tell us who will use the resource.
We use verified-email access to protect SPECS intellectual property and maintain a proportionate record of professional requests.
- Names and professional roles accept letters and standard punctuation.
- Country is selected from a controlled list.
- Email and telephone numbers are checked for valid structure.
- Marketing permission is optional and separate from access.