Your Universal Learning Assistant
An AI-native operating system for thought — with persistent memory, agentic RAG, and proactive intelligence.
YULA isn't just another chatbot. It's an AI operating system that remembers everything, learns your patterns, and proactively helps you before you even ask.
┌─────────────────────────────────────────────────────────────┐
│ │
│ "Remember that restaurant I mentioned last month?" │
│ │
│ YULA instantly retrieves the context, knows your │
│ preferences, and suggests a reservation. │
│ │
└─────────────────────────────────────────────────────────────┘
Intelligent routing that decides when to search your memory — not just how.
// Fast route: Pattern matching (no LLM call, <1ms)
"Hi" → direct_reply
// LLM route: Complex decisions via Groq Llama (<100ms)
"What did we discuss about React hooks?" → rag_searchEvery conversation becomes searchable knowledge. Vector embeddings via Qdrant enable semantic search across your entire history.
Proactive AI Callbacks — YULA doesn't just respond, it reaches out when it matters.
┌────────────────────────────────────────┐
│ 🔔 PAC Notification │
│ │
│ "You mentioned wanting to review │
│ your project proposal today. │
│ Ready when you are." │
│ │
└────────────────────────────────────────┘
Unified interface to the best models, each chosen for their strengths:
| Provider | Model | Use Case |
|---|---|---|
| Groq | Llama 3.1 | Fast routing decisions |
| Anthropic | Claude 3.5 | Complex reasoning & code |
| OpenAI | GPT-4o | General assistance |
yula/
├── apps/
│ ├── web/ # Next.js 15 + React 19
│ └── mobile/ # React Native (Expo)
├── services/
│ └── api/ # Hono API server
├── packages/
│ └── db/ # Prisma + PostgreSQL
└── lib/
└── ai/ # Vercel AI SDK layer
├── router.ts # Agentic RAG router
├── providers.ts # Multi-provider registry
└── generate.ts # Unified generation
|
Frontend
|
Backend
|
|
AI Layer
|
Infrastructure
|
# Clone
git clone https://github.com/EfeDurmaz16/aspendos-deploy.git
cd aspendos-deploy
# Install
bun install
# Setup database
bun db:generate
bun db:push
# Environment
cp .env.example .env.local
# Add your API keys
# Run
bun dev# AI Providers
OPENAI_API_KEY=sk-...
ANTHROPIC_API_KEY=sk-ant-...
GROQ_API_KEY=gsk_...
# Database
DATABASE_URL=postgresql://...
# Vector Store
QDRANT_URL=https://...
QDRANT_API_KEY=...
# Auth
BETTER_AUTH_SECRET=...
BETTER_AUTH_URL=https://yula.devYULA embraces a bold, minimal design language:
| Element | Value |
|---|---|
| Background | #050505 / #000000 |
| Accent | Electric Amber #F59E0B |
| Border | #FFFFFF10 |
| Font | Geist / Geist Mono |
No purple-blue gradients. No visual noise. Just signal.
POST /api/chat/stream
Content-Type: application/json
{
"messages": [{ "role": "user", "content": "Hello" }],
"model": "gpt-4o"
}POST /api/memory/search
Content-Type: application/json
{
"query": "React hooks discussion",
"limit": 5
}GET /api/health
→ { "status": "healthy", "uptime": 12345 }We welcome contributions! Please see DEPLOYMENT.md for setup instructions.
# Run tests
bun test
# Type check
bun run typecheck
# Lint
bun biome check .MIT © 2024 YULA OS
yula.dev · Built with Vercel AI SDK
Your thoughts, amplified.