ORI 2.0 - AI Orientation Companion

L'Étudiant's ORI AI already generated 50,000 conversations and 570,000 messages with a 24% login rate — but operated as a stateless Q&A engine with no student profile, no memory, no progression. France's official ratio: 1 guidance counselor per 1,000 students.

How to transform ORI from a conversational search engine into a personalized decision companion — guiding a high school student from 'I don't know what to do' to an argued shortlist of programs, salon preparation, and tailored application documents?

ORI's real Vertex AI Reasoning Engine API, L'Étudiant editorial content via RAG (1,500+ articles/year, 120 salons, tens of thousands of indexed programs), and student profile data collected during onboarding.

Built a fully functional prototype in React 18 + TypeScript + Zustand connected to ORI's real API. Designed 'Le Miroir' — a 60-second rich visual onboarding with interactive world map, illustrated school level selection, and thematic affinity cards — replacing 5-8 minutes of text Q&A. Built a real-time orientation tree (Profil → Domains → Programs → Schools) auto-updating during conversation, with SVG custom renderer at 0 API tokens. Added Voxtral voice interface with French TTS and a CV + cover letter generator with 3 configurable tones and in-chat AI editor. Mapped 8 product extension directions with an impact/feasibility matrix and built a B2C/B2B revenue model.

Won the Prix Coup de Coeur at the Alberthon. Delivered a live prototype connected to L'Étudiant's real ORI API, demonstrating a full product vision shift: not a better chatbot, but a decision product with visual memory.

Reframed ORI's market position: one private counselor session costs €80–150; ORI Premium at €24.99/year serves an entire cycle. Built a full go-to-market with freemium, family, and B2B institutional tiers (~€8/student/year for high schools) — anchoring the product as an accessibility play, not a luxury tool.