Reads Handwritten Papers
Accepts PDF, Word or photos of answer sheets, reads each page with a vision model and corrects crooked scans automatically.
An AI-powered product that reads scanned or photographed exam answer sheets, marks each answer against the teacher's own reference material, and draws the marks back onto the student's handwriting — with a human examiner always in control.
Upload the student's answer paper along with the question paper, reference material and marking scheme — QIQ Paper Checker handles the rest, and the examiner reviews every mark.
Accepts PDF, Word or photos of answer sheets, reads each page with a vision model and corrects crooked scans automatically.
Reads the paper's structure, marks and rubric — including "answer any 3 of the following 5" — so students are scored out of what was required, not what was printed.
Answers are evaluated against the teacher's own textbook, notes and model answers — not the model's memory.
Marks and annotations are drawn back onto the student's own writing, with an annotated transcript as a fallback.
Every mark is a proposal, never a decision. Examiners can override anything and the totals, grade and report update instantly.
A printable report with score, grade, per-question breakdown, positives and improvement areas for every student.
The full application is live — upload an answer paper, watch the AI mark it, and review every mark as an examiner. The source is open on GitHub for developers.
A browser application with no heavy dependencies — retrieval, geometry and segmentation are hand-written.
API keys never reach the browser; a server-side proxy schedules and protects them.
A detection failure is never reported as a student failure, and a failed evaluation is never a zero.
Hundreds of assertions pin the behaviour of matching, marking, choice rules and geometry.
We focus on measurable value, human oversight and products that can be operated and improved over time.