ClassBuddy
Ethical AI platform for social-emotional support of primary school students through Conversational AI.
Focus/Ethical AI for education
Stage/Pilot and active research
The Challenge
Many children hesitate to talk to adults about sensitive, emotional topics such as bullying, anxiety, or friendships. This leads to missed signals for teachers and unprocessed emotions in children.
The Solution
Based on the founder's vision, I built the complete dual-sided platform as lead developer:
AI-Chatbuddy (Student)
A safe, judgment-free space where children can anonymously chat with specialized AI "Buddies" (e.g., the 'Bullying Buddy' or 'Anxiety Buddy') about their feelings and problems. This is a low-threshold first step for emotional expression.
Teacher Dashboard (Teaching Staff)
An ethical dashboard that gives teachers insight into social and emotional trends at the class level without violating the child's privacy.
Technical Implementation
The core of ClassBuddy lies in the balance between help and privacy:
Child Privacy
Chats are GDPR-compliant. Conversations are stored exclusively as metadata (about which topic was discussed). Teachers can never read the actual conversation content.
Teacher Insight
Teachers see trends (e.g., "Five students discussed 'Bullying' this week"). This serves as an early warning system, allowing them to provide timely human guidance.
Successful Pilot: We deployed a successful pilot at a primary school. Within one week, this resulted in more than 1,500 voluntary chat sessions, proving strong product-market fit and the urgency of the need.
Traction & Status
Active Research: After the pilot, I joined as co-founder. The platform is currently a working product that we are actively using as a team for in-depth research at multiple primary schools.
Portfolio Value
This project demonstrates my expertise in:
Ethical AI & Security
Developing AI for sensitive use cases with Privacy-by-Design (including robust backend security and strict GDPR protocols).
Full-Stack Development
From user-facing (React app) to professional dashboard (data visualization).
Concept-to-Impact
The ability to transform a complex ethical concept into a scalable, functional product with demonstrable value for real users.
Built with a modern stack
Frontend
- Next.js
- React
- TypeScript
- Tailwind CSS
Backend
- Supabase (PostgreSQL, Auth, Edge Functions, Realtime)
AI
- OpenAI API
Tools
- Vercel