Contract Closed

TrueMark: Developing a Teacher-Controlled AI Writing Platform

NCER

Program: Small Business Innovation Research

Award amount: $250,000

Project director: William Riley

Awardee:

TrueMark Systems Corp.

Year: 2025

Award period: 8 months (09/12/2025 - 05/11/2026)

Project type:

Phase I Development

Contract number: 91990025C0079

Purpose

TrueMark's existing platform enables teachers to administer AI-enhanced writing assignments with transparency and control. This Phase IB project develops a personalization engine that learns from teacher grading patterns to adapt AI tutoring and assessment for individual students. The new component automatically analyzes how teachers edit rubric scores and comments to identify student-specific weaknesses, then uses these insights to personalize the AI copilot and rubric checker. After three graded assignments, students receive guidance tailored to their unique writing challenges based on their teacher's past feedback. Teachers maintain full control through a dashboard where they can view and modify student profiles. This approach transforms teacher grading—an activity educators already perform—into the data source that powers individualized instruction without requiring teachers to manually configure settings or become prompt engineers.

Project Activities

The project develops three integrated components. First, a data processing pipeline that batch-processes teacher feedback after each assignment, comparing teacher edits against AI suggestions to identify class-wide and student-specific patterns. Second, a teacher dashboard that displays system-detected weaknesses and allows educators to edit student profiles before they become active. Third, personalized AI interactions where the copilot and rubric checker reference student profiles to provide tailored guidance.

Development runs October through December 2025, with a single prototype release on January 15, 2026. The pilot study involves 10 teachers and 200 students across three partner schools, running through March 2026. Students complete multiple writing assignments using the personalized AI features while the research team collects quantitative data on system usage, AI engagement patterns, and alignment between AI suggestions and teacher grading. Qualitative data includes teacher interviews, classroom observations, and pre/post student surveys. Analysis in April-May 2026 examines whether the personalization engine improves the relevance and usefulness of AI feedback compared to generic AI tutoring tools.