sykang16/agentic-wealth-intelligence

1---2title: Agentic Wealth Intelligence3emoji: 📈4colorFrom: blue5colorTo: purple6sdk: streamlit7sdk_version: 1.55.08app_file: ui/streamlit_app.py9pinned: false10---11 12# Agentic Wealth Intelligence13 14AI-powered wealth management system with multi-agent orchestration, conversational profiling, RAG-enhanced recommendations, and real-time market data integration.15 16## Architecture17 18**LangGraph Hybrid Supervisor Routing** — two-tier intent classification feeds into a flat outer graph; the Recommend node embeds a Supervisor subgraph for multi-step context gathering.19 20### Outer Orchestrator Graph21 22```23                 ┌──────────────────────────────────────┐24     User ──────►│             Router                   │25                 │  Tier 1: Keyword match  (fast path)  │26                 │  Tier 2: LLM fallback   (T=0)        │27                 └──────────────┬───────────────────────┘28                                │ conditional routing29       ┌────────────────────────┼──────────────────┬──────────────┐30       ▼                        ▼                  ▼              ▼31┌────────────┐       ┌──────────────┐   ┌──────────────┐  ┌──────────┐32│ Portfolio  │       │  Profiling   │   │  Recommend   │  │ General  │33│  Module A  │       │   Module B   │   │  [Subgraph]  │  │  (LLM)   │34│ AssetAgent │       │ slot-filling │   │  Supervisor  │  │          │35└──────┬─────┘       └──────┬───────┘   └──────┬───────┘  └────┬─────┘36       └────────────────────┴──────────────────┴────────────────┘37                                                │38                                         ┌──────▼──────┐39                                         │   Respond   │──► END40                                         └─────────────┘41```42 43### Recommendation Supervisor Subgraph44 45```46┌───────────────────────────────────────────────────────────────┐47│             Supervisor  (LLM decision · guard: steps <= 5)    │48│        ┌────────────────────┬──────────────────┐              │49│        ▼                    ▼                  ▼              │50│ ┌─────────────┐   ┌──────────────┐   ┌──────────────┐        │51│ │portfolio_   │   │profiling_    │   │  recommend   │        │52│ │fetch        │   │fetch         │   │  synthesis   │        │53│ │AssetAgent   │   │get_profile_  │   │  RAG + MCP   │        │54│ │.process()   │   │  summary()   │   │              │        │55│ └──────┬──────┘   └──────┬───────┘   └──────┬───────┘        │56│        └─────────────────┴──────────────────┘                 │57│                    loops back to Supervisor                    │58│                                          finish ─────► Respond │59└───────────────────────────────────────────────────────────────┘60```61 62**Modules:**63- **Portfolio Analysis** — Natural language queries about your holdings, net worth, allocation, gains/losses64- **Investment Profiling** — Conversational slot-filling to build your risk/preference profile65- **Recommendations** — AI-generated investment advice using portfolio data, RAG knowledge base, and live market data66- **AI Advisor** — Unified chat interface that automatically routes to the right module67 68## Quick Start69 70### 1. Install Dependencies71 72```bash73pip install -r requirements.txt74```75 76### 2. Configure Environment77 78Copy `.env.example` to `.env` and set your API keys:79 80```bash81cp .env.example .env82```83 84Required (at least one LLM provider):85- `OPENAI_API_KEY` — OpenAI GPT-4o86- `ANTHROPIC_API_KEY` — Anthropic Claude87- `GEMINI_API_KEY` — Google Gemini88 89Optional (for live data):90- `ALPHA_VANTAGE_API_KEY` — Market quotes and financial news91- `NEWS_API_KEY` — Business news92- `SEC_USER_AGENT` — SEC EDGAR filings93 94### 3. Generate Synthetic Data95 96```python97from backend.src.data_generation import generate_sample_data98generate_sample_data()99```100 101### 4. Run the Streamlit UI102 103```bash104python -m streamlit run ui/streamlit_app.py105```106 107### 5. Run the API Server108 109```bash110uvicorn backend.src.api.app:app --reload111```112 113API documentation available at `http://localhost:8000/docs`.114 115## API Endpoints116 117| Method | Endpoint | Description |118|--------|----------|-------------|119| `POST` | `/api/v1/chat` | Unified orchestrator chat |120| `GET` | `/api/v1/portfolio/{user_id}` | Portfolio summary |121| `POST` | `/api/v1/portfolio/{user_id}/query` | Natural language portfolio query |122| `POST` | `/api/v1/profiling/{user_id}/start` | Start profiling session |123| `POST` | `/api/v1/profiling/{user_id}/respond` | Continue profiling conversation |124| `POST` | `/api/v1/recommendations/generate` | Generate recommendations |125| `GET` | `/health` | Health check |126 127## Project Structure128 129```130backend/131├── src/132│   ├── agents/              # Agent implementations133│   │   ├── asset_agent.py   # Portfolio query agent134│   │   └── orchestrator.py  # Multi-agent orchestrator135│   ├── api/                 # FastAPI application136│   │   ├── app.py           # App factory137│   │   ├── dependencies.py  # Dependency injection138│   │   ├── schemas.py       # Request/response models139│   │   └── routes/          # API route handlers140│   ├── asset_management/    # Module A: Portfolio data141│   ├── common/              # Shared models, LLM client142│   ├── multi_agent/         # LangGraph orchestrator143│   │   ├── state.py         # State definitions144│   │   ├── routing.py       # Intent classification145│   │   ├── nodes.py         # Graph node functions146│   │   └── graph.py         # Graph construction147│   ├── profiling/           # Module B: Slot-filling agent148│   ├── recommendation/      # Module C: RAG + recommendations149│   └── mcp/                 # MCP servers for live data150ui/151├── streamlit_app.py         # Main Streamlit application152tests/153├── unit/                    # Unit tests154├── integration/             # Integration tests155└── e2e/                     # End-to-end tests156```157 158## Running Tests159 160```bash161# All tests162pytest tests/ -v163 164# Specific test categories165pytest tests/unit/ -v166pytest tests/integration/ -v167pytest tests/e2e/ -v168```169 170## Tech Stack171 172- **Python 3.11+** — Core language173- **LangGraph** — Multi-agent orchestration and profiling workflows174- **FastAPI** — REST API layer175- **Streamlit** — Interactive UI176- **ChromaDB** — Vector store for RAG177- **Sentence Transformers** — Document embeddings178- **MCP (Model Context Protocol)** — Live market data integration179- **Pydantic v2** — Data validation180 181## Documentation182 183See [docs/00_INDEX.md](docs/00_INDEX.md) for the full documentation index.184