Multi Agent Financial Assistant

Links:

GitHub

Overview:
The Financial Assistant Agent is a multi-agent AI system designed to assist users with comprehensive financial inquiries. From generic conceptual discussions and continued conversations to highly specific price-action analysis (e.g., "What happened to NVDA in the last couple of days?"), the platform provides data-grounded and concise insights.

Built on a modular architecture powered by LangGraph and FastAPI, the platform intelligently combines specialized autonomous agents to process technical, fundamental, and sentiment-based signals. The system is designed for multi-turn interactions, allowing users to follow up on complex findings with further questions.

The system is architected to support an Agent-to-Agent (A2A) communication pattern (v2.0), where a central Advisor Agent will orchestrate tasks between specialized units like the Technical Analysis Agent.

Key Capabilities:

• 🤖 Conversational Intelligence: Handles both conceptual finance queries and real-time market deep-dives with full context awareness.
• 📈 Data-Grounded Analysis: Multi-agent collaboration to synthesize Technical, Fundamental, and News/Sentiment indicators.
• 📉 Future-Ready (A2A): Planned autonomous delegation between the central Advisor and specialized Market Intelligence units.
• 🔌 MCP Integration: Uses the Model Context Protocol (MCP) for extensible tool discovery and execution.
• 📊 Agent Tracing: Observability powered by MLflow for tracking agent trajectories, prompt versioning, and cost-performance trade-offs.

🚀 Key Features:

• Intelligent Routing: Automatically determines whether a query requires conceptual explanation or data-driven analysis.
• Multi-Turn Workflows: Supports continued conversations, allowing users to drill down into specific data points or broader market trends.
• LangGraph State Management: Maintains conversational context across complex reasoning loops and conditional edge transitions.
• Asynchronous Execution: High-concurrency backend designed for parallel tool calls and multi-agent interaction.
• Concise Outputs: Specialized prompts ensure data is prioritized over fluff, providing clear and actionable financial summaries.

🏗️ Architecture:
The project follows a Multi-Agent Network design:

1. User Request hits the Advisor Agent (e.g., "Why is MSFT trending up?").
2. Advisor Agent analyzes intent and delegates sub-tasks (Technical, News, Fundamentals) to the specialized units.
3. MIU (Technical Agent) uses LangGraph to pull live indicators (SMA, RSI, etc.) via MCP Tools.
4. Knowledge Synthesis: Sub-agent results are synthesized into a unified, concise response back to the Advisor.
5. Final Delivery: User receives a grounded answer and can initiate follow-up questions immediately.