Reflection on the Process

Borde – Da Silva – Lizy Pousson – Goujon | Licence 3 – IRT - 2025/2026

ÒCTAVIA is an idea we have been refining for several years, so its core direction did not fundamentally change during Phase 2. What did change — significantly — is which parts of the project we put at the front, and how we frame what makes ÒCTAVIA actually distinct. Three moments drove that evolution.

Refocusing on Cognitive Load Reduction as Our Core Differentiator

Reducing the student's mental load through the Planner agent has always been central to our vision. Yet, while building the POC, we naturally drifted toward the more visible features — the Socratic agent, RAG, multi-modal content generation — and let the Planner slip into the background. Dr. Alain Goudey's keynote, together with comments comparing our project to tools like NotebookLM, forced us to confront a sharp question: "What actually makes your project an innovation? What makes it unique?" After internal discussion, we put the organizational dimension and mental-load reduction back at the very center of the project. This is what separates ÒCTAVIA from a "smart course assistant": we treat the student as a whole person under pressure, not just a learner asking content questions.

Rethinking Model Choice — and Adding an Ecological Dimension

Discussions with our mentor and as a team led us to formalize our model strategy. We have requested a partnership with Mistral AI to scale efficiently, but we also concluded that ÒCTAVIA must remain functional with smaller public models such as Llama. Rather than treating this as a fallback, we turned it into a design principle: optimizing the system for smaller models reduces energy consumption regardless of which model is ultimately deployed. This adds an explicit ecological dimension to a project that was previously framed only around sovereignty and pedagogy.

Confirming the Human-Centered Direction

Saiph Savage's keynote on Human-Centered AI resonated directly with our intent to put the student back at the center of their own learning journey. It did not redirect us, but it gave us sharper language and stronger conviction for what we were already doing: every component of ÒCTAVIA is built around the student, not the reverse.

Overall, the Challenge process strengthened our proposal less by changing what we build, and more by clarifying why we build it that way — and which features carry that "why" most visibly.