Flowdeck is an AI-powered stock analysis platform built for independent investors. It combines real-time market data, multi-agent AI analysis, and a conversational Trader Copilot — so you can research stocks, read full AI reports, and ask follow-up questions all in one place.
| Feature | Description |
|---|---|
| Live Market Data | Real-time prices, volume, bid/ask, 52-week range, and today's range for any ticker |
| Multi-Agent AI Analysis | A committee of specialized AI agents (market, news, fundamentals, technical, sentiment) collaborates to produce a BUY / SELL / HOLD recommendation with full reasoning |
| Transparent Reports | Six report tabs per stock — Market, News, Fundamentals, Technical, Sentiment, Investment Plan — so you can read the full AI reasoning, not just the final label |
| Trader Copilot | A side-by-side workspace: stock detail panel on the left, AI chat on the right. Ask follow-up questions about any stock while viewing its live data and reports |
| AI Analyst Agent | Standalone conversational AI with live access to prices, fundamentals, news, technical indicators, insider activity, and your saved reports. Responses stream in real-time |
| On-Demand Reports | Generate fresh AI analysis for any stock, any time — using your token balance |
| Stock Subscriptions & Dashboard | Subscribe to stocks and receive email updates when new analysis reports are available. Track your watchlist from a personalized dashboard |
| Token Economy | New users receive 1,000 free tokens on sign-up. Additional tokens can be purchased in packs |
flowdeck/
├── ai_engine/
│ ├── tradingagents/ # Multi-agent analysis graph (analysts, researchers, risk mgmt, trader)
│ └── watchlist_consulting/ # Watchlist deep-research pipeline
├── backend/ # FastAPI API server
│ ├── main.py
│ ├── models/ # Pydantic + DB models
│ └── services/ # Business logic (analysis, chat, reports, tokens, email…)
└── frontend/ # React + TypeScript app (Vite)
└── src/
├── pages/ # Full-page views (Home, Stock, Copilot, Chat, Dashboard…)
├── components/ # Reusable UI components
└── services/ # API clients and type definitions
1. Clone and install
git clone <your-repo-url>
cd flowdeck
python -m venv venv && source venv/bin/activate
# or: conda create -n flowdeck python=3.11 && conda activate flowdeck
pip install -r requirements.txt2. Configure API keys
Copy the example env file and fill in your keys:
cp backend/.env.example backend/.envRequired keys:
OPENAI_API_KEY— or Anthropic / Google / Azure; seeai_engine/tradingagents/default_config.pyALPHA_VANTAGE_API_KEY— for fundamentals and news (free key)
3. Run backend and frontend
# Terminal 1 — API server (http://localhost:8002)
cd backend && python main.py
# Terminal 2 — Frontend dev server (http://localhost:3003)
cd frontend && npm install && npm run devOr use the convenience scripts:
./scripts/start_flowdeck.sh # start both
./scripts/stop_flowdeck.sh # stop bothThe Copilot (/copilot) is a three-panel workspace designed for active research:
┌─────────────────┬──────────────────────────────┬──────────────────┐
│ Ticker List │ Stock Detail Panel │ AI Copilot Chat │
│ (watchlist) │ live quote · report tabs │ streaming · live│
│ │ fundamentals · news │ data access │
└─────────────────┴──────────────────────────────┴──────────────────┘
- Left panel — your watchlist; add any ticker with the search box
- Middle panel — full stock detail: live quote, AI recommendation, all report tabs, fundamentals, news, insider activity
- Right panel — AI chat with full context: the selected ticker, all watchlist tickers, live prices, reports, and more
The Copilot chat is context-aware: when you select a ticker, the AI automatically knows which stock you're looking at and can answer questions like "What are the key risks for NVDA?" or "Compare MSFT and GOOGL fundamentals" without you having to specify the ticker each time.
The standalone AI Analyst Agent (/chat) gives you a full-page conversational interface with the same capabilities as the Copilot chat:
- Live stock quotes and historical data
- AI-generated reports and BUY/SELL/HOLD recommendations
- Fundamentals, balance sheet, cash flow, income statement
- Technical indicators
- Insider transactions and sentiment
- Company and global news
- Your watchlist and portfolio context
Responses stream token-by-token. Each message shows the number of tools accessed and tokens consumed, so you always know what data the AI used.
Flowdeck's analysis is produced by the tradingagents package — a LangGraph-based multi-agent graph. You can run it directly in your own scripts:
from ai_engine.tradingagents.graph.trading_graph import TradingAgentsGraph
from ai_engine.tradingagents.default_config import DEFAULT_CONFIG
config = DEFAULT_CONFIG.copy()
# config["deep_think_llm"] = "gpt-4o"
# config["quick_think_llm"] = "gpt-4o-mini"
ta = TradingAgentsGraph(debug=True, config=config)
_, decision = ta.propagate("NVDA", "2024-05-10")
print(decision)Analysis flow:
- Specialized analysts — Market, News, Fundamentals, Technical, and Sentiment analysts each produce a focused report and score
- Bull vs. Bear debate — Two AI researchers argue both sides across multiple rounds, stress-testing the investment case
- Research manager + trader — Synthesizes the debate into an investment plan and actionable stance
- Risk debate → recommendation — Aggressive, cautious, and neutral risk perspectives debate before a final judge produces the BUY / SELL / HOLD
Configuration options: ai_engine/tradingagents/default_config.py
| Doc | Description |
|---|---|
| docs/DEPLOYMENT.md | Production deployment (systemd, Nginx, SSL) |
| docs/DEPLOYMENT_UBUNTU.md | Ubuntu-specific deployment guide |
| docs/DATABASE_MIGRATION.md | Running database migrations |
| docs/STOCK_DEEP_RESEARCH.md | Stock Deep Research agent (comprehensive company reports) |
| docs/AI_ANALYSIS_FLOW.md | Detailed AI analysis pipeline documentation |
| docs/BILLING_INTEGRATION_SUMMARY.md | Token economy and PayPal billing |
See LICENSE.