TOC Meeting Notes - 2025-09-10

1 Attendees
2 Meeting Opening and Attendance
3 2026 Conference Planning
4 TOC Elections Update
5 WGU Team Progress Report
6 OpenAI Funding Opportunity
7 AI Experimentation Framework Presentation
8 Sparkth Tool Demonstration

Attendees

The Technical Oversight Committee met on Sep 10, 2025 with the following members in attendance:

Anant Agarwal (2U)
Aref Matin (2U)
Dustin Tingley (Harvard)
Edward Zarecor (Axim)
Nacho Despujol (Instructor Representative, UPV)
Régis Behmo (Operator & Core Contributor Representative, Edly)
Xavier Antoviaque (User/Learner representative, OpenCraft)

Guest presenters:

Felipe Montoya - Co-founder and CTO of Edunext, presented the AI experimentation framework

Meeting Opening and Attendance

Several regular TOC members were unable to attend due to scheduling conflicts, including Jeremy and Ferdi who had a conflict with Open Learning. It was noted that Shweta Gupta had left NSDC and the TOC with immediate effect, and that her replacement for NSDC is pending.

2026 Conference Planning

The 2026 Open edX conference location and host have been confirmed and will be announced. There is also a plan to co-locate future conferences with Democon, and the next step is to discuss the coordination of planning efforts.

TOC Elections Update

The nomination process for TOC elections is currently underway, with the elections now following a biennial schedule rather than annual. Steps have been taken to ensure a smoother voting process following last year’s change of voting tool and the issues it generated.

WGU Team Progress Report

WGU's new team supporting Open edX has demonstrated impressive onboarding and contribution levels. The team has submitted 300-400 pull requests and now has three engineers with commit rights to the platform. This rapid integration and high contribution volume was noted as particularly noteworthy for a new institutional participant.

Discussion touched on WGU's work with student profiles and learner records, with mention of concepts around comprehensive learner profiles that could inform personalized learning experiences.

OpenAI Funding Opportunity

Axim is exploring ways to access an OpenAI innovation funding opportunity, despite exceeding the typical operating budget cap that would normally qualify organizations. Potential approaches through partnerships or other mechanisms are being investigated to access this funding for AI-related development work.

AI Experimentation Framework Presentation

Felipe Montoya, Co-founder and CTO of Edinext, presented a comprehensive AI experimentation framework designed to integrate with Open EDX. The framework builds upon the existing Hooks framework and aims to be pluggable, model/provider agnostic, and well-integrated with Open EDX services. A GitHub repository called "open-edx-ai-extensions" has been created to house this work, with basic backend and frontend functionality already available in the main branch.

The current proof of concept focuses on a simple "Ask AI" button for content summarization, though it was acknowledged to be intentionally over-engineered to provide maximum extensibility for future use cases. The framework addresses current challenges where AI integrations have low adoption rates, with existing solutions often being custom-built under deadline pressure. The hope is that this standardized framework will lower the barrier to entry and encourage wider experimentation across the Open EDX community.

Discussion revealed that Axim plans to make significant investments in this AI experimentation framework over the next couple of quarters, indicating strong institutional support for the initiative.

Sparkth Tool Demonstration

Regis demonstrated Sparkth, an innovative external, open-source MCP (Model Context Protocol) server that enables course generation and management within Large Language Models like Claude. This tool works with both Open edX and Canvas platforms, leveraging existing APIs without requiring any modifications to the core platform code.

The demonstration showed how users can interact with Claude to generate complete courses, including content creation and direct publishing to Open edX instances. The tool represents a complementary approach to AI integration, working externally rather than as an embedded framework. Sparkth is fully open-source and available for community use and contribution.

Questions were raised about the tool's capabilities with multimedia content and different course formats, with clarification provided about current limitations and potential future enhancements.