Multi-Agent AI Finance Assistant: Open Source Financial Analysis Platform with Multi-Agent Collaboration

This post introduces an open-source financial analysis platform based on a multi-agent AI framework—Multi-Agent AI Finance Assistant. The platform integrates large language models (e.g., Google Gemini) with professional financial algorithms. Through the collaborative work of eight intelligent agents, it provides functions like stock research, market prediction, portfolio optimization, and risk assessment. It covers over 460 global stocks and cryptocurrencies, supports local/cloud deployment, and helps users efficiently gain financial insights.

With the development of AI technology, the digital transformation of the financial industry is accelerating. Traditional financial analysis tools have single functions and isolated data, making it difficult to meet modern investors' needs for real-time, comprehensive, and intelligent analysis. The Multi-Agent AI Finance Assistant project emerged as the times require, integrating LLM and financial algorithms through a multi-agent collaborative architecture to build a powerful open-source platform.

The platform adopts a modular multi-agent architecture, decomposing complex tasks into 8 professional agents: API Agent (data acquisition), Scraping Agent (news crawling), Retriever Agent (historical data retrieval), Analysis Agent (in-depth analysis), Language Agent (natural language processing), Prediction Agent (return prediction), Graphing Agent (visualization), Voice Agent (voice interaction). Clear division of labor improves efficiency and accuracy.

The platform supports over 460 global stocks, covering U.S. stocks (NASDAQ, NYSE), Asia (South Korea, Japan, Hong Kong, India), Europe (UK, France, Germany, Switzerland), Canada and Australia markets, as well as 98 ETFs and more than 20 cryptocurrencies. All data is obtained in real-time via the Yahoo Finance API to ensure analysis is based on the latest market information.

Core functions include: 1. Intelligent stock screening (industry classification, custom search); 2. Portfolio risk assessment (correlation, volatility calculation); 3. Machine learning return prediction (polynomial regression + visualization comparison); 4. Real-time news sentiment analysis; 5. Rich visualization (stock price comparison, trading volume analysis, etc.); 6. Automated report generation (integrating data and analysis results).

Tech Stack: Frontend Streamlit, Backend FastAPI, AI Model Google Gemini, Data Source Yahoo Finance API, Visualization Matplotlib/Plotly. Deployment Methods: 1. Local Deployment (recommended, no cloud restrictions): Start FastAPI backend + Streamlit frontend; 2. Cloud Deployment (Streamlit Cloud/Render, note free plan limitations). Version v2.0.0 fixes dependency conflicts, optimizes code synchronization, etc.

Target Users: Individual investors, financial analysts, developers/researchers, educational institutions. Value: Lower analysis barriers, improve decision-making efficiency, optimize risk control, provide learning resources. The project uses the MIT license, encourages community contributions, has a modular architecture for easy function expansion, and has gained attention on GitHub.