Human-in-the-loop AI grading

Grade essays alongside an AI agent.

GraderJet reads the paper, scores it against your rubric, and updates the workspace live — while you stay in control of every deduction, score change, and curve.

Start grading

No account needed — works with or without an API key.

Agentic grading

The agent calls tools as it writes — updating rubric scores, flagging passages, and curving the batch. The workspace reacts in real time.

Grade a whole class

Paste or drag & drop .txt, .docx, and .pdf essays to build a batch. Navigate papers with ⌘← / ⌘→ and curve them together.

Human in the loop

Every change is auditable. The agent explains each deduction, and you adjust, override, or ask for a re-read at any point.

Insights & exports

Class summary with distributions and category averages — export the batch as CSV or hand back student-ready feedback reports.

Comment bank

Notes you write become reusable one-click snippets, so the fifth paper's feedback is as fast as the first.

Private & offline

Progress autosaves in your browser — no account, no server-side storage. No API key? A mock agent demos the full flow.

How it works

1

Paste the paper

Enter the student name, the essay text, and the assignment prompt — or load a sample essay.

2

The agent assesses

Rubric scores and feedback flags appear on the document as the agent works through the paper.

3

Interrogate & adjust

“Why did Evidence lose two points?” Ask, override, or curve the batch — everything is logged.

Ready to grade your first paper?

Set up a paper