You mark. It drafts. You finalise.
Marking Assist turns your grade and a few notes into feedback in your own style, written against your rubric. You edit it, then sign it off. Nothing reaches a student until you finalise it.
- Students’ scripts deleted when you finalise
- No subscription
Brief overview — what it is and why it matters
Plays from YouTube when you press play; YouTube (Google) may set cookies then. See the cookie notice.
Your notes come before the AI draft
Tools that show you a machine’s grade first invite you to accept it. Marking Assist asks for your grade and your judgements before it writes a word, then drafts feedback that explains them. The draft never marks the work; it never changes what you wrote.
How it works
Six steps, from a new assignment to exported marks. The AI writes only at step four, and only from what you decided.
- Create the assignment
Step 1: Add the brief, the rubric and any marking guide or graded sample. Preview five feedback styles made from them and keep the one that sounds like you.
- Upload the scripts
Step 2: PDF, Word, PowerPoint, Excel, images or text. Each is shown as the pages the student made, with a permanent script number and, if you use them, a Student ID.
- Read, grade and write your notes
Step 3: Mark as you always have: read the script beside your rubric, give the grade or band, and jot your judgements. Your comment bank is one key away.
- The AI drafts from your marking
Step 4: Only after you have marked, Gemini writes feedback that explains the grade you gave, from your notes, your rubric and your chosen style. It never grades the work and never writes from the script alone.
- Edit, and compare with the draft
Step 5: Change anything in place. Every version is kept, so you can compare your edit with the draft and ask for a second draft with a note on what to change.
- Finalise and export the marks
Step 6: Sign the feedback off in your name. The student's file is deleted at that moment. Export the marks as CSV and the feedback as text files for your university's system.
The full walkthrough
A narrated tour from a new assignment to finalised feedback: context files, a feedback style, uploading scripts, the workspace and the comment bank, a draft, editing, comparing and signing off.
From assignment to finalised feedbackComing soon
A narrated tour of the whole workflow, with captions
How much time would it give back?
Put in your own numbers. The reading stays the same; the writing is what changes.
Marking by hand, AI alone, or Marking Assist
What each one asks of you, and what each one leaves behind.
| What | By hand | AI alone | Marking Assist |
|---|---|---|---|
| Who decides the grade | You | The model | You. The draft explains the grade you gave. |
| Who writes the feedback | You, from a blank page | The model, from the script alone | The model drafts from your notes, rubric and style; you edit. |
| Who signs it off | You | Nobody | You. Nothing reaches a student until you finalise. |
| Consistency across a cohort | Varies between markers and over a long run | The same model every time, unchecked | Every draft uses your rubric and style; you and your moderation decide. |
| A record of what changed | Only if you keep one | Usually none | Every version kept; compare any edit with the draft. |
| What happens to the student's file | Stays on your machine | Depends on the tool | Stored in the EU, sent to Gemini only to draft, deleted when you finalise. |
What the research says
Four things the peer-reviewed evidence supports, and where Marking Assist sits in each. Every citation links to its source.
Feedback helps when it informs
Not all feedback is equal: comments that explain how to improve outperform rewards or bare right-or-wrong corrections.
Across 435 studies, feedback rich in information had the largest effect on learning (d = 0.99); rewards and punishment the smallest (d = 0.24) (Wisniewski et al., 2020).
d = 0.99 vs 0.24High-information feedback against rewards or punishment (meta-analysis of 435 studies, school to university).
In Marking Assist
Marking Assist drafts rubric-aligned comments that the marker reviews and edits before release, so marking time can go on specific, actionable feedback.Marking is hard, human work
Time pressure and repetition are real constraints, and early evidence suggests they can affect judgements, not just wellbeing.
In a UK experiment, boredom rose across ten essays and higher boredom went with lower marks; a rubric did not prevent it (Erturk et al., 2022).
In Marking Assist
Marking Assist shows the marker, and sends the AI, each script as rendered pages: scripts and slides as the pages the student made, and tables as cleaned text, so judgements rest on the work as the student submitted it.Criteria help, but judgement varies
Rubrics improve transparency and learning; they do not remove the tacit judgement at the heart of marking.
24 experienced UK assessors varied in how they chose, ranked and scored criteria; the authors call that variability "inevitable" (Bloxham et al., 2016).
g = 0.45The effect of rubrics on academic performance (meta-analysis of 21 controlled studies, school to university).
In Marking Assist
Marking Assist drafts feedback against your own rubric criteria, while calibration, moderation and the final judgement stay with your marking team.AI assists; humans should decide
Evidence on AI feedback is promising but mixed; on AI grading of complex work, it is weaker still.
90.5% of 6,960 students rated teacher feedback trustworthy, against 60.1% for AI feedback (Henderson et al., 2025).
60.1% vs 90.5%Students who rated AI feedback, against teacher feedback, as trustworthy (n = 6,960, four Australian universities).
In Marking Assist
Marking Assist drafts feedback for the marker to check, and the marker makes the final grading decision; none of the research cited here evaluated Marking Assist.
Studies cited evaluate feedback, marking and AI in general; none evaluated Marking Assist. Read the full evidence review with references.
What you get
Your materials first
Upload the brief, rubric, marking guide and graded samples once per assignment; every draft is written against them.
Every file type
PDF, Word, PowerPoint, Excel, images and text, read as the pages the student made.
A feedback style of your own
Preview feedback styles on one of your real scripts and keep the one that sounds like you.
A comment bank that comes to you
312 starter comments by criterion and band, plus your own, one key away while you write notes.
Student IDs and anonymous marking
Mark by ID or by script number, with file names hidden from you and from the AI.
Insights per assignment
Grade and band distributions, and a review that reads your finalised feedback for consistency.
Marks out, feedback out
Export marks as CSV and each student's feedback as a text file named by their ID.
Credits, not a subscription
One credit drafts one script; the first regeneration is free. Credits last 12 months.
Three colours, one meaning each
Everywhere in Marking Assist, amber is what you wrote, blue is what the AI drafted, and green is what you signed off. You can always see who wrote what, and so can anyone who asks.
- Amber: your notes. Private, never changed by the AI.
- Blue: the draft. Always labelled, always waiting for you.
- Green: finalised. Only you can make something green.
Built for how universities assess
Marking Assist fits the way assessment already works in UK higher education, rather than asking you to change it.
Academic judgement stays with the marker
You grade first. The AI writes up your judgement; it never replaces it.
Marked against the published criteria
Every draft is written against the rubric and brief you uploaded, in the words of your marking scheme.
Consistent across a cohort, ready for moderation
The same style, the same criteria and the same structure for every script, with each draft read and edited by you.
A record of every version
Each draft, each edit and the final sign-off is kept, for moderation, appeals and external examiners.
Feedback students can act on
Specific comments tied to criteria, with what to do next, in a tone you chose.
This reflects the expectations set out in the UK Quality Code for Higher Education for assessment that is valid, reliable and transparent, with academic judgement exercised by staff. No framework or regulator has reviewed or endorsed Marking Assist.
Pricing
Starter
£6
25 credits · 24p a script
For a single assignment or a first try.
Module
£12
60 credits · 20p a script
For a module's marking across a term.
Term
£30
200 credits · 15p a script
For several modules or a heavy marking season.
No subscription. One credit drafts one script; the first regeneration is free, then each costs half; failed drafts are refunded.
Paid securely with Stripe. Primoga Ltd is not registered for VAT. No VAT is charged. Credits last 12 months from the day they are added. The oldest are used first.
Departments and institutions: contact us for a quote and invoice.
Buying for a department? Your data-protection officer will want the data-protection summary: where data is held, who processes it and when files are deleted.
Questions
Does the AI mark the work?
No. You give the grade and your judgements; the AI drafts the wording. Nothing is finalised without you.
Who sees the students' scripts?
Only you. A script is sent to Gemini only while its draft is written, and the file is deleted when you finalise.
What does it cost?
Credits, bought in packs; one credit drafts one script. Every account starts with 2 free credits.
What happens to a student's file?
It is stored in the EU, read once to draft, and deleted when you finalise, when you delete the script, or 30 days after upload at the latest. Each deletion is logged.
Can I mark on a phone?
Yes. The workspace fits a phone, with the script and your marking as two tabs, and the comment bank as a sheet.
Your judgement, written up faster
Create an account, upload one assignment’s scripts and draft two for free. If it does not sound like you, you have lost nothing.
Create an account