The AI builds a full AI grader for a teacher grading 100+ papers a week
Liang Gou wrote no code himself: he directed the AI to design and build a complete grading app for his wife, a high school math teacher grading 100+ papers a week at 3 to 5 hours per test. The rubric engine reconstructs a rubric from a single graded example, its eval bank comes from real AP Calculus AB free-response exams, two models cross-check grades, and the agent tested the app itself in a browser before deployment.
How it works
- Direct the AI to design the grader and write a design doc
- Agent builds a Flask app whose rubric engine reconstructs rubrics from a graded example or AP-style format
- Agent builds an eval bank from real AP Calculus AB free-response exams
- GPT and the AI cross-review each grade
- Agent tests the app itself in a Chromium browser
- Deploy on Render with a Neon database
- Teacher grades her first class with it
Why it’s worth knowing
He wrote zero code: the agent designed, built, browser-tested and deployed the grader, and built its exam bank from real AP Calculus questions.
Where it’s from
Summarized from a public post on lianggou.github.io. See the original ↗