AI feedback for AP & IB writing.
AP and IB students hand in essays and wait days for paragraph-level feedback that's often a single rubric score and a sentence. Teachers don't have the hours, and existing AI graders treat all writing the same — generic prose tips against generic rubrics. Nothing was actually scoring against the rubrics that determine the grade.
We built an agentic pipeline that runs multiple passes against the real College Board / IB rubrics — one pass per criterion, then a synthesizer. Students paste an essay, pick a subject, and get a rubric-locked score plus paragraph-level feedback in under a minute. The product surface stays minimal so the feedback is the thing you read; everything else fades.
Live with thousands of students across the US and abroad, used heavily during exam season. Recently completed a content-asset acquisition with OnePrep / RevisionDojo (YC) — transferred the practice-test library and MCQ pipeline; kept the platform, brand, codebase, and user base. Provisional patent on the agentic essay-grading pipeline.
More subjects, slower. The win is rubric-fidelity, not coverage — adding an exam means months of rubric work and teacher review.