β¨ Inspiration
Mental health support and relationship guidance often lack accessibility and personalization. Moh AI Mate was inspired by the need for an AI-driven, empathetic conversational agent that can offer emotional support and relationship advice in a human-like manner.
π₯ What It Does
Moh AI Mate is a next-gen AI assistant that:
βοΈ Engages in human-like conversations using a fine-tuned LLM
βοΈ Understands emotions and offers personalized guidance
βοΈ Provides relationship advice with a therapeutic approach
βοΈ Remembers past interactions for seamless, context-aware discussions
βοΈ Works across platforms (Web & Mobile)
π οΈ How We Built It
Tech Stack:
- Frontend: React.js (ShadCN UI, TailwindCSS, Framer Motion)
- Backend: Flask (Python) with REST APIs
- AI Models: Fine-tuned DeepSeek, Mistral, LLaMA 2
- Mobile: React Native
- Deployment: Vercel, Railway, Render
π We integrated memory persistence, allowing Moh AI Mate to retain context for meaningful, personalized conversations.
π§ Challenges We Ran Into
- Ensuring natural & emotionally intelligent responses from AI.
- Handling memory & long-term user context efficiently.
- Optimizing real-time interaction speed while keeping conversations dynamic.
- Frontend animations & UI/UX complexity.
π Accomplishments That Weβre Proud Of
βοΈ Successfully developed a therapeutic AI chatbot
βοΈ Implemented emotion-aware conversation memory
βοΈ Designed a modern UI/UX that feels engaging and responsive
βοΈ Integrated multiple LLMs to compare performance
π What We Learned
- Fine-tuning LLMs for specific emotional intelligence tasks.
- Optimizing real-time AI-generated conversations.
- Handling memory persistence for long-term user interactions.
- Creating an intuitive chat UI with animations.
π Whatβs Next for Moh AI Mate
π Voice & Speech Input (AI-powered voice conversation)
π Multi-Agent AI System (more specialized AI personalities)
π Personalized Therapy Plans (AI-driven well-being programs)
π Mobile App Release for Android & iOS
π Offline AI Chatbot (Run Moh AI locally with on-device models)
Log in or sign up for Devpost to join the conversation.