Suggestions with evidence
Rubric items come with the page and the expression they rest on, so you check a suggestion against the page instead of re-grading it.
Early access · engineering courses
Grade handwritten STEM homework with an assistant that shows its work.
Partial Credit reads scanned and photographed solutions, suggests rubric items with the page evidence behind them, checks the math exactly where it can, and never releases a grade until you approve it.
Students: use the course code from your instructor. Instructors and students can sign in with an email link or Google.
2 State feedback 8 / 10
0Correct gain derivation
−2Sign slip in the characteristic polynomialsuggested
−4Controllability not justified
Exact check: closed-loop poles match the target
Awaiting your approval
How it works
- 1
Set up the assignment
Upload the problem and your key, and build a Gradescope-style rubric of signed items. Open it to your course with due and late-due times.
- 2
Students submit
A PDF or phone photos, mapped page by page to questions. They can resubmit until the deadline; late work is marked, and every attempt is kept.
- 3
Review, approve, publish
Open each question on its own pages, accept or change the suggested rubric items, and approve. Students see grades only after you publish.
Built for how engineering is graded
Exact math checks
For configured controls, linear-algebra and statics questions, answers are verified exactly. Unverified full credit is labeled, not hidden.
Question-by-question review
Click a question and see only its mapped pages, like Gradescope. Handwriting, typed work and photos all live in one queue.
Courses and rosters
Import a roster CSV or share a course code limited to your school’s email. TAs grade; only instructors publish.
Deadlines, extensions, regrades
Due and late-due times, per-student extensions, and regrade requests answered from one queue with the full history.
Measured, not promised
Import your own Gradescope grades and see exactly where the AI agrees with you, where it is lenient, and which rubric items it misses.
Principles
- You approve every grade. AI output is a draft. Nothing reaches a student until an instructor approves and publishes it.
- Student work stays student work. It is never used to train AI models, and you can permanently delete any submission.
- Honest about uncertainty. Unreadable work is flagged, not guessed; verified and unverified results are labeled differently.
Ready when you are.
Sign in to your courses, or join one with your instructor’s code.