Socrates — AI That Teaches You How to Think
Inspiration
AI tools are making coding faster—but also more passive.
We saw people copy-paste solutions without truly understanding them.
So we asked:
What if AI guided thinking instead of replacing it?
What it does
Socrates is an AI coding mentor that:
- Breaks problems into steps
- Guides users with hints (not answers)
- Analyzes code and explains mistakes
- Encourages learning through interaction
How we built it
- Frontend: Next.js + Tailwind + Framer Motion (cinematic UI)
- Backend: Serverless API routes on Vercel
- AI: Google Gemini API for step generation, hints, and code analysis
Flow:
User → Steps → Code → Feedback → Improvement
Key Innovation
Instead of:
$$ \text{Output} = f(\text{Input}) $$
We built:
$$ \text{Learning} = f(\text{Input}, \text{Interaction}, \text{Reflection}) $$
Challenges
- Balancing guidance vs giving answers
- Designing prompts that teach, not solve
- Optimizing serverless performance
Accomplishments
- Built a true AI mentor, not a chatbot
- Created a learning-first interaction model
- Designed a cinematic, immersive UI
What we learned
- AI should amplify thinking, not replace it
- Prompt engineering is critical
- UX directly impacts learning
What’s next
- Adaptive learning system
- Voice-based mentor
- Personalized AI learning styles
“I cannot teach anybody anything, I can only make them think.” — Socrates
Built With
- api
- css
- gemini
- javascript
- next.js
- react
- tailwind
- typescript
- vercel
Log in or sign up for Devpost to join the conversation.