Development Roadmap
Our planned timeline for building and deploying the Open Knowledge Tracing Service.
Knowledge Tracing Algorithm Finalization
Complete development of the unified declarative-procedural Knowledge Tracing methodology and validate performance on existing conversational KT datasets.
MCP Server & ChatGPT App SDK Development
Build the core infrastructure including authenticated endpoints, Google OAuth integration, and ChatGPT Apps interface.
System Integration & Testing
End-to-end integration of all components with comprehensive internal testing before public release.
Public Deployment
Launch on ChatGPT Apps marketplace and begin initial user onboarding.
Data Collection & Iteration
Longitudinal data collection, monitoring user engagement, and iterative refinement of the declarative-procedural transition model.
Open Source Release
Open-source the service, enable local deployment, and release the research dataset for the academic community.
Success Metrics
Pre-Launch Validation
Offline predictive performance comparable to established Knowledge Tracing benchmarks through evaluation against baseline models (e.g., AFM-based KT) using existing educational datasets.
User Adoption
Learning Interactions
Target: ≥200,000 total interactions (initial), with stretch goals at 500,000+. Includes problem attempts and conversational tutoring turns.
Learning Gain
Positive change in predicted mastery over time, measured through pre-tests and post-tests aligned to Knowledge Components and normalized by estimated difficulty.
Long-term Sustainability
After initial project resources are exhausted, the service can be sustained at minimal cost (~$450/year for Supabase and Vercel). The system will be open-sourced, allowing users to download their data and run the service locally or deploy independently.