Traditional platforms store your data.
Mentron acts on it.
Learning management systems were designed for an era of file uploads and manual record-keeping. They faithfully record that a student scored 41% and then wait for someone to notice. Meanwhile teachers absorb the grading load, and students receive instruction calibrated to the middle of the class.
Mentron treats the gap between assessment and intervention as the product. It evaluates work against your rubrics, identifies the specific misconception, generates material that targets it, and measures whether mastery actually moved — with a teacher approving every mark along the way.
The loop, stage by stage
Four stages, each producing the input for the next. The fourth stage is what makes it a loop rather than a pipeline.
Observe
AI evaluation of graded work extracts structured misconception tags — not a sentiment score, but specific, machine-readable errors such as sign_error or unit_conversion.
Target
Those tags are injected directly into generation prompts, so the flashcards, mindmap or notes a student receives address the exact gap rather than the topic in general.
Measure
Mentron records whether the student opened the material and what happened to their score on the next attempt — the mastery delta, per topic.
Adapt
Effectiveness is stored per tool and topic pair. Future dispatches shift toward whatever demonstrably moved mastery for that topic, so the loop improves with use.
Four things that are different
The closed loop
Most AI learning tools generate content and stop. Mentron treats generation as the middle of the process, not the end. Misconceptions are extracted from real graded work, targeted material is dispatched automatically, and the effect on mastery is measured and fed back into tool selection.
- Misconception tags extracted from graded submissions
- Study artifacts generated against those specific tags
- Return-and-improve measured as a mastery delta
- Tool choice biased by what worked for that topic
Approval-gated grading
AI evaluates each submission criterion by criterion against your rubric and drafts per-student feedback. The result is held in a pending state — invisible to the student, absent from the gradebook, excluded from analytics — until a teacher reviews, edits and approves it.
- Rubric-aligned, criterion-level evaluation
- Nothing released without explicit teacher approval
- Edit before approving; the teacher owns the mark
- Approval writes back to mastery and analytics
Outcomes and accreditation
Course outcomes are structured data across storage, generation and analytics — not free text in a syllabus. Every generated question carries its outcome and Bloom level, so attainment reporting is a query rather than a spreadsheet assembly job before each inspection.
- Bloom K1–K6 tagging on generated questions
- Course outcome arrays across every pipeline
- Live CO × Bloom coverage heatmaps
- KHDA/MoE 70/50 Group A/B attainment logic
Deployment without a project plan
Evaluation happens in a pre-seeded sandbox with synthetic institutional data — no setup, no student records, no procurement prerequisite. Adoption starts by ingesting materials and rosters you already have.
- Day-one ingestion from Google Drive, Classroom and Canvas
- Pre-seeded sandbox for risk-free evaluation
- Multi-tenant isolation and role-based access control
- Regional hosting options for data residency
Built for every role in the institution
A platform only survives if the people who have to use it daily get something out of it.
For teachers
Reclaim the evenings
- Generate quizzes, assignments and full exam papers from materials you already have
- Review AI-drafted rubric evaluations instead of authoring feedback from scratch
- Get a ranked reteach list each week — which topics, which students, before the exam
- Keep final authority over every mark that reaches a student
For heads of department & leadership
Evidence, not assembly
- Attainment dashboards across departments, courses and cohorts, computed continuously
- Mastery velocity to identify at-risk cohorts weeks ahead of assessment
- Cross-course skill gaps that reveal foundational problems, not isolated symptoms
- Inspection-ready outcome reporting without a spreadsheet cycle
For students
Study what you actually got wrong
- Flashcards, mindmaps and notes generated from your own performance gaps
- Ask questions against your course materials and get cited answers
- Spaced repetition queues that schedule review at the right interval
- Guided hints from an AI tutor grounded in the course, not the open internet
Common questions
See it running, not described
The sandbox is a full synthetic institution — rosters, submissions and multi-week history already in place. Walk the teacher, student and department flows end to end. Access is provisioned per email address, so start with a short request.
No student data required. The sandbox runs entirely on synthetic records.