Graditas

Grade written exams faster — you approve every result.

Graditas reads scanned handwritten exams, drafts scores against your rubric, and builds a gradebook you review, override, and approve. Nothing is final until you say so.

How it works

Step 1

Upload your materials

A rubric (Word, PDF, or YAML), your class roster, and the scanned exams as a single PDF batch.

Step 2

AI drafts a gradebook

Each answer is read from the scan and scored against your rubric criteria, with partial credit and per-question reasoning you can inspect.

Step 3

You review and approve

Open any student, check the reasoning, override any score. The gradebook stays a draft until you approve it — then export CSV.

AI-assisted grading — faculty always approve. Graditas never finalizes a grade on its own.

Built for FERPA-sensitive work

Student data stays in AWS

OCR and grading run entirely inside AWS (Textract and Bedrock). Student work is never sent to third-party AI providers.

Graded without names

Written answers are graded under anonymous aliases and re-identified only in your gradebook.

Encrypted at rest

Uploads, working files, and results are stored in S3 with server-side encryption (SSE-KMS).

Audit trail

Account and grading actions are recorded in an audit log.

Built for university faculty and TAs

Handwritten exams, quizzes, and problem sets — graded against your rubric, reviewed by you.