Avilon - AI Therapy Bot POC

A minimal proof-of-concept AI therapy chatbot using Next.js, Supabase, and OpenRouter.

Features

• Intake Wizard: 5-question assessment including PHQ-2 and GAD-2 screening
• AI Chat: Context-aware therapy conversations using CBT techniques
• Two Therapy Modes:
  • Quick check-in (unstructured chat)
  • Guided CBT exercises (thought challenging, deep breathing, grounding)
• Crisis Detection: Automatic keyword detection with crisis resources
• Dashboard: Session history, mood trends, and progress tracking
• Memory System: Stores last 10 messages + patient profile for context

Tech Stack

• Frontend: Next.js 14 (App Router), React, TypeScript, Tailwind CSS, shadcn/ui
• Backend: Next.js API Routes
• Database: Supabase PostgreSQL with pgvector
• AI: OpenRouter (deepseek/deepseek-r1-distill-llama-70b at $0.27/1M tokens)
• Auth: NextAuth.js with Supabase
• Deployment: Vercel (free tier)

Setup Instructions

1. Clone the repository

git clone <repository-url>
cd avilonai.com

2. Install dependencies

npm install

3. Set up Supabase

1. Create a project at https://supabase.com
2. Go to SQL Editor and execute the schema from lib/db/schema.sql
3. Copy your project URL and keys from Settings > API

4. Set up OpenRouter

1. Create an account at https://openrouter.ai
2. Generate an API key from the dashboard
3. Add credits to your account ($5-10 is sufficient for POC testing)

5. Configure environment variables

Copy .env.example to .env and fill in your credentials:

cp .env.example .env

Required variables:

• NEXT_PUBLIC_SUPABASE_URL: Your Supabase project URL
• NEXT_PUBLIC_SUPABASE_ANON_KEY: Supabase anonymous key
• SUPABASE_SERVICE_ROLE_KEY: Supabase service role key (for admin operations)
• NEXTAUTH_URL: Your app URL (http://localhost:3000 for development)
• NEXTAUTH_SECRET: Random secret (generate with openssl rand -base64 32)
• OPENROUTER_API_KEY: Your OpenRouter API key
• NEXT_PUBLIC_APP_URL: Your app URL

6. Run the development server

npm run dev

Visit http://localhost:3000

Deployment to Vercel

1. Push your code to GitHub
2. Import project to Vercel
3. Add environment variables in Vercel dashboard
4. Deploy

Usage Flow

1. Sign Up: Create an account at /signup
2. Complete Intake: Fill out the 5-step intake wizard
3. Start Session: Choose session type and rate your mood
4. Chat: Engage with Avilon using CBT techniques
5. Dashboard: View your session history and mood trends

Crisis Detection

The system automatically detects crisis keywords and provides immediate resources:

• 988 Suicide & Crisis Lifeline
• Crisis Text Line (741741)
• Emergency room guidance

All incidents are logged for safety monitoring.

Cost Estimate

• Infrastructure: $0 (Vercel + Supabase free tiers)
• LLM Costs: ~$5-10/month for 100-200 conversations
• Total: Under $10/month for POC

Project Structure

/app
  /api
    /auth          # Authentication endpoints
    /chat          # Chat and session endpoints
    /intake        # Intake form submission
    /dashboard     # Dashboard data
  /(auth)          # Login and signup pages
  /(dashboard)     # Protected dashboard pages
/components
  /chat            # Chat interface
  /intake          # Intake wizard
  /ui              # shadcn/ui components
/lib
  /ai              # AI agent configuration
  /db              # Database schema and client
  /utils           # Utilities (crisis detection, etc.)

Security & Privacy

• User authentication via NextAuth.js + Supabase
• Environment variables for API keys
• Server-side API routes for sensitive operations
• Crisis incidents logged for safety

Important Notes

⚠️ This is a POC for educational purposes only.

• Not a replacement for professional mental health care
• Not HIPAA compliant
• Not suitable for production use without additional security and compliance measures
• Crisis detection is keyword-based and may have false positives/negatives

License

MIT

Support

For issues or questions, please open a GitHub issue.