Penn Carey Law AI Project  /  Build

Build

Tools that put research into practice — software, platforms, and pipelines.

The Project’s build pillar produces working software: virtual TAs, exam-creation tools, AI-grading pipelines, judicial chambers tools, course-material repositories, simulation environments. Default is open source.

Code lives at github.com/pennai-law.

Projects in Build

11
001

Exam Grader

Calibration-based AI grading for law-school essay exams — deployed against Polk's Spring 2026 IP class. buildresearch Active
002

Course Materials

An extraction primitive: turn the messy artifacts of a real course — PDFs, slide decks, casebook excerpts, syllabi — into clean structured text that downstream tools can use. buildteach Active
003

Casebook Builder

Curate and edit cases into custom course casebooks at a fraction of the commercial cost — faculty pilots in Torts and Patents (Fall 2026) and Legislation (Spring 2027). buildteach Active
004

Exam Taker

The AI Final Exam Project — can current AI pass real Penn Carey Law finals when its work is graded blind, on the curve, by the faculty who wrote the exams? Wave 1 results are in. researchbuild Active
005

Course-Website Tooling

Generalize the Legal Education at Model Velocity microsite into reusable templates the Project and others can deploy quickly. buildteach Planning
006

Essay Creator

Generate assessment-grade essay exams — issue spotters and fact-pattern questions — with rubrics built for AI-assisted grading. buildteach Active
007

MCQ Creator

Generate law-school multiple-choice questions with built-in distractor validation and psychometric quality controls. buildteach Active
008

Heron — A Course-Bounded Virtual TA

A virtual teaching-assistant chatbot bound to one course's materials: it answers students in Slack from the assigned readings, with citations. Plus the argument that faculty should consider building these to govern the AI students already use. Spring 2026 results and a draft paper. buildteach Active
009

Class Simulations

AI-driven simulations built around the actual class problems law students work through each week — starting with patent law. buildresearch Planning
010

Newsletter Automation

Build the pipeline that produces the Project's monthly AI newsletter — automated collection, editorial workflow, Substack distribution. buildteach Active
011

Training Materials

Produce a complete AI training package for legal-education audiences — and pitch a cross-Penn version with collaborators. teachbuild Active