01
Create assignment
A lecturer sets the assignment and test configuration.
Submission comprehension assurance
AI can write the assignment. Can the learner defend it?
Sentinel™ gives learners a short, submission-specific comprehension test immediately after they submit their work. Instructors gain a practical signal for deciding who may require further oral review.
Not AI detection. Evidence of understanding.
*Click on image to view larger
The problem
A polished submission no longer guarantees that learning happened.
Generative AI can produce convincing academic work quickly. At the same time, AI-detection claims remain unreliable, blanket bans are difficult to enforce, and full oral examination for every learner is too expensive for most courses.
Sentinel™ helps instructors screen the whole class and concentrate further investigation where it is most needed.
How Sentinel™ works
A five-step workflow designed for real teaching environments.
Sentinel™ is a signal, not a verdict. Questions are based on the learner’s own submission and results help instructors decide where oral follow-up may be worthwhile.
02
Learner submits work
The platform receives the learner’s specific essay, report, or draft.
03
Questions generated
Sentinel™ produces short multiple-choice, true/false, and short-response checks.
04
Timed assessment completed
Learners complete the submission-linked comprehension assessment immediately.
05
Selective oral follow-up
Instructors review scores, explanations, and evidence to choose cases for further discussion.
The JITA™ engine
Turn the text that matters into an assessment that matters.
Just-in-Time Assessment uses the learner’s or class’s actual material to create focused questions at the moment understanding needs to be checked.
Instead of a generic quiz bank, JITA™ identifies the key ideas in a submission, passage, or study document and builds a targeted assessment around them.
1 · Source text
Submission, passage, or notes
2 · JITA™
Key ideas become claims
conceptsevidencereasoning
3 · Assessment
A focused understanding check
✓ Multiple choice
✓ True / false
✓ Short response
The result: practical evidence of understanding for Sentinel, Reading Tests, and Study Mode.
Three JITA™ applications
Support understanding at every stage of learning.
Sentinel
Written-assignment assurance: test whether learners understand the work they submit.
Reading Tests
Develop class comprehension with instructor-led passages and assessments.
Study Mode
Help learners sharpen recall from their own study material.
For instructors
Screen the whole class. Investigate selectively.
- Quickly identify learners with weak demonstrated understanding.
- Reserve oral examination for failed or anomalous cases.
- Review questions, answers, scores, and supporting evidence.
- Re-mark attempts when a generated question is defective.
- Reduce the burden of manually examining every learner.
For learners
Check understanding before submission.
- Identify concepts that require further study.
- Improve confidence in defending submitted work.
- Receive feedback grounded in the learner’s own document.
- Treat assessment as part of learning rather than punishment.
- Practise privately before the official submission event.
Clarification
Sentinel™ does not try to prove who wrote the work.
Not an AI detector
Sentinel™ does not claim to determine whether a submission was AI-generated.
Not a plagiarism checker
It is not a substitute for similarity analysis or existing integrity workflows.
Not a replacement for judgment
Instructors remain responsible for interpreting the signal and deciding what follow-up is proportionate.
Sentinel™ provides structured evidence of whether a learner understands the work they submitted.
Pilot design
Help shape the first version of Sentinel™.
We are inviting a small group of instructors and academic programmes to test a barebones MVP in real teaching environments. Early adopters will be selected based on assignment suitability, class size, subject area, willingness to provide structured feedback, and willingness to conduct limited oral follow-up for validation.
Early pilots will evaluate question quality, lecturer usefulness, agreement with oral follow-up, learner experience, and willingness to adopt the tool repeatedly.
Who this is for first
- University lecturers and college instructors
- Programme leaders and teaching offices
- Assessment-heavy subjects with defendable written work
- Teams willing to validate outcomes with limited oral review
FAQ
Questions institutions are likely to ask first.
Does Sentinel™ detect AI-generated work?
No. Sentinel™ does not attempt to prove authorship or detect AI generation. It assesses whether learners can demonstrate understanding of what they submitted.
Are learners prevented from using AI?
No. The premise is different: learners may use AI if permitted, but they should still be able to explain and defend the submitted work.
What happens if a generated question is poor?
Instructors can review questions and re-mark or adjust outcomes if a generated item is defective. The MVP is designed to support judgment, not replace it.
How can learners prepare independently?
Study Mode lets learners turn their own notes and material into recall practice. It is separate from assignment assurance and class Reading Tests.
Does Sentinel™ replace oral examinations?
No. It helps reserve oral follow-up for the subset of cases where extra verification appears justified.
Which assignment formats are supported first?
The MVP is best suited to essays, reports, case studies, and similar text-based submissions. Format coverage can expand based on pilot feedback.
Early access
Apply for early access.
We will review applications and invite suitable pilot participants in stages. Acceptance is selective and depends on course context, assignment design, and pilot fit.