DeepMarking for institutions
For unit coordinators, schools and faculties who want to try AI-assisted marking on a real assessment, with the answers your IT and privacy teams will ask for.
Who it is for
- Unit coordinators with large cohorts, several markers and a rubric that has to be applied the same way by everyone.
- Schools and faculties looking at AI-assisted marking for more than one unit, and needing a clear account of where student work goes.
- Learning and teaching teams running a trial before a wider decision.
Students never use DeepMarking. Markers upload the work, review the marks and export them to your LMS.
What changes for a team
Institutions use the Enterprise plan. It is set up with us rather than bought online.
- No plan caps. Enterprise accounts have no limit on marking credits, active assessments or cohort size.
- Limits set per account. We can set assessment, cohort and credit limits for each marker, for example to fit a pilot.
- Each marker has their own account. Projects, submissions and marks belong to the account that created them and are not visible to other markers.
- The same tools as every plan. Your rubric, group work, review of every mark, and exports to Excel, PDF and LMS files. See Features.
If your team needs something not listed here, tell us when you get in touch.
Data, privacy and security
Short answers to the questions IT and privacy teams ask first. The Privacy Policy and Terms of Service are the full versions.
Where is our data stored?
Accounts, projects, rubrics, the text extracted from submissions, and marks are stored in DeepMarking’s Supabase database. Our providers may process data outside your country, for example in the United States or the EU and UK. Ask us for the current hosting regions before a pilot.
Which AI model marks the work?
Google Gemini, through DeepMarking’s own platform account. The rubric, the specification, the text and images of each submission, and the student name and ID read from the file names are sent to Gemini to produce marks and feedback. Lecturers cannot connect their own AI keys.
Is student work used to train AI models?
DeepMarking does not train or fine-tune models on your marking. When a lecturer confirms or edits marks, short examples are added to later prompts for the same assessment so marking follows their preferences; student ID numbers and labelled name lines are stripped from those examples. How Google may use requests sent to Gemini depends on Google’s terms for the account we use, and our Privacy Policy makes no further promise. Ask us for the current terms before a pilot.
Who controls the student data?
The lecturer or institution that uploads student work is normally the controller of that data, and needs permission under its own policies to upload it. DeepMarking processes it to provide the service and does not sell student personal data.
How long is data kept, and can it be deleted?
When a batch starts, each student’s extracted work waits in private storage until it is marked, then it is deleted; leftovers from a batch that never finishes are deleted within a day or two. The text a submission was marked from is kept with its marks. Lecturers can delete single projects, their marking history or their whole account, which removes the profile, projects, submissions and marks. Residual backups and logs may persist for a limited period.
What goes into product analytics?
Analytics use identifiers, counts and events, never student names or submission text, and session replay masks student work. Each lecturer can turn analytics off in their settings.
Who are the processors?
Supabase for sign-in and the database, Google for Gemini, Stripe for payments, PostHog for product analytics, an email provider for account emails, and the host that serves deepmarking.com.
How a pilot works
- 1
Pick one unit and one assessment
Choose an assessment with a clear rubric and a cohort big enough to show the time saved. The coordinator and one or two markers are enough.
- 2
Mark alongside your usual process
Mark part of the cohort in DeepMarking while markers work as they normally do. Markers review, confirm or change every mark before anything is released.
- 3
Compare the results
Look at where markers changed the AI marks, how long review took, and whether the exported sheets fit your LMS.
- 4
Widen it
Add more assessments, units or markers, with plan limits set to the cohort sizes you need.
Talk to us about a pilot
Tell us the unit, the assessment and roughly how many students. Please do not send student work in the message.