Setting Up an Exam Batch and Tuning Auto-Approval

A batch is a container for all papers from a single exam sitting. Organising papers into batches lets you track class-level statistics, compare individual students against the cohort, and export a whole class's grades at once.

Go to Batches → New Batch. Fill in: exam name, subject, class/year, date, and total marks. You can add an optional description — useful for distinguishing 'Mid-term 2026' from 'Re-sit Mid-term 2026'.

Open the batch and go to its Question Paper / Answer Key editor. At the top, "Generate from an existing paper" lets you drag in a scan or PDF, snap a photo directly from your phone (no QR code needed if you're already on the phone), or paste raw text — the AI reads it and builds out the questions, sections, sub-parts, and MCQ options automatically, matching the structure of the paper you gave it.

Every generated question shows a plain editable text field for its prompt and its expected answer, with a live rendered preview appearing underneath any field that contains math notation — so you can immediately confirm equations came through correctly. You can also add or edit questions by hand at any point, whether or not you generated from a document first.

Under Settings, the auto-approval threshold controls which papers are approved automatically vs held for review — it's a percentage (0-100) applied to a paper's average confidence across all its questions. The default is 88: a paper whose average confidence is at or above that is approved automatically; anything below lands in your review queue.

For a class test where speed matters, lower this. For board-exam preparation or important internal assessments, raise it — or keep it high so more papers get a human check before results go out.

Once papers start flowing in, the batch detail page shows a live score distribution chart, class average, highest and lowest marks, and per-question average — so you can spot which question the whole class struggled with before you've finished grading.