Chat with Fundamentals is an application that generates executive summaries about stocks using fundamental data fetched via EODHD APIs. It integrates financial metrics, historical stock prices, recent company news, and AI-written executive summaries into a single orchestrated workflow — designed to streamline the fundamental analysis of securities.
Built with CrewAI, LangChain, FastAPI, and EODHD APIs market data.
Chat with Fundamentals is not intended to be a SaaS platform. It is a research engine designed to explore how autonomous agents and AI can automate and enhance financial analysis. It is built to be self-hosted and tailored to user needs through the refinement and upgrade of AI and quantitative workflows. All LLM interactions are logged and saved for reproducibility.
Key principles:
- Modularity – Workflows and agents are designed independently
- Reproducibility – All steps (query, fetch, summarize) are transparent and testable
- Minimal Overhead – Lightweight backend with no complex infrastructure
- User Ownership – Users bring their own API keys (OpenAI, EODHD, Tavily)
backend/
├── agents/ # CrewAI agents (e.g., fundamental interpreter)
├── core/ # Configuration, logging, LLM settings
├── models/ # Pydantic models for finance data
├── routers/ # API routes (analyzer, simulater, quantanalyzer)
├── tools/ # EODHD API tools (data, news fetching)
├── utils/ # Helper functions (e.g., schema generation)
├── workflows/ # CrewAI-based orchestration (e.g., analyze.py)
├── venv/ # Virtual environment (optional)
├── .env # Environment variables (API keys, model names)
├── .gitignore
└── chatwithfundamentals.log
frontend/
├── components/ # Reusable React components (UI, layout, forms)
├── Icons/ # Custom icon components
├── lib/ # Client-side utilities and API routes
├── pages/ # Next.js pages
├── public/ # Static assets (images, fonts)
├── styles/ # Global and modular CSS/Tailwind styles
├── types/ # TypeScript type declarations
├── LICENSE.md # Project license information
├── NOTICE.txt # Legal and attribution notices
├── next.config.js # Next.js configuration
├── next-env.d.ts # Next.js environment type definitions
├── package.json # Project metadata and dependencies
├── package-lock.json # Exact dependency tree
├── postcss.config.js # PostCSS configuration (used with Tailwind)
├── tailwind.config.js # Tailwind CSS configuration
└── tsconfig.json # TypeScript compiler configuration
Chat with Fundamentals can be run locally on Linux/macOS or Windows. Setup scripts are included for convenience.
bash deploy.shdeploy.batThese scripts will:
- Create a Python virtual environment
- Install Python and frontend (Node.js) dependencies
- Copy
.env.modelto.envif it does not exist
Once installed, launch the full application stack with a single command.
bash launch.shlaunch.batThese launch scripts will:
- Open the project in VS Code
- Start the frontend (Next.js) at http://localhost:3000
- Start the backend (FastAPI) at http://localhost:8000
- Open both interfaces in the default browser
Required file: AAPL.US.json
Do not delete or rename this file. It is used as a reference knowledge base for agents.
You may duplicate and modify it for other tickers — but the original must remain.
Example Query: "Is TSLA a good buy?"
Workflow:
-
Parses the query and identifies TSLA as the target ticker
-
Fetches OHLCV, fundamentals, and news via EODHD APIs
-
Generates a concise LLM-written executive summary
-
Saves raw data in structured JSON
-
Provides extended analytics:
- TradingView-style chart
- Monte Carlo simulation
- VaR estimation
- Returns histogram
- Beta and correlation vs benchmark
- 3-year cumulative return comparison
As of version 1.0.1, all tickers must include their market suffix (e.g., .US, .PA). For example, use AAPL.US instead of AAPL, or AIR.PA instead of AIR.
Example Query: "Compare TSLA and AMZN"
Workflow:
-
Recognizes both tickers
-
Fetches data and news for each
-
Constructs comparative summary:
- Financial performance
- Price and sentiment
-
Displays side-by-side analytics
As of version 1.0.1, all tickers must include their market suffix (e.g., .US, .PA). For example, use AAPL.US instead of AAPL, or AIR.PA instead of AIR.
Example Query: "What are the last articles about AI and Quant Finance from ArXiv"
Workflow:
- Performs recursive search with GPT-Researcher
- Builds a structured research report
- Provides source links
- Monte Carlo Simulation
- Value at Risk (VaR) and Expected Shortfall
- Daily return histograms
- Beta and R² plots
- 3-year cumulative return comparison
Planned features include:
- Rolling Sharpe ratios
- Calendar heatmaps
- ACF/PACF plots
- Volatility cones
- Drawdown visualizations
pip install -r requirements.txtOPENAI_API_KEY='your_openai_key_here'
EODHD_API_KEY='your_eodhd_key_here'
TAVILY_API_KEY='your_tavily_key_here' (optional)uvicorn main:app --reloadfrom backend.workflows.analyze import FundamentalFlow
flow = FundamentalFlow()
result = flow.invoke(inputs={"user_query": "Analyze AAPL and TSLA fundamentals and forecast returns"})
print(result.model_dump_json(indent=2))- Intraday data support
- Alphalens integration
- News sentiment analytics
This project is intended for research and personal development use. Commercial deployments must comply with OpenAI,EODHD and GPT-Researcher licensing.
Current Version: v1.0.1
- Added GPT-Researcher support
- Volatility and performance ratios
- In development branch
Baseline Version: v1.0
- Agentic workflow foundation
- Multi-ticker comparison
- Includes 4 of 10 planned quant modules
- Stock scanner to be added in QuantAgents
This application and its associated research outputs are provided solely for informational and educational purposes. Nothing contained within this repository, its documentation, or its outputs should be construed as financial advice, or as an offer, recommendation, or solicitation to buy or sell any security, product, or service.
Users are solely responsible for their own investment and trading decisions.
The author(s) make no representations or warranties as to the accuracy, completeness, or suitability of the content for any purpose.
By using this application, you acknowledge that you must exercise independent judgment, conduct your own due diligence, and refer to the Terms of Use and Disclaimer pages provided within the application.
S.M. Laignel
Founder of SL MAR consultancy
AI and Quantitative Finance
GitHub: sl-mar/chat-with-fundamentals