FinWell: Your Personal Finance & Health Copilot on Fetch.ai

FinWell is a decentralized AI copilot that helps users manage stocks, crypto, and health coverage—powered by Fetch.ai’s uAgents, ChatProtocol, and ASI1 Mini LLM.

Comment

FinWell: Your Personal Finance & Wellness Copilot

About the Project

What Inspired Us

The inspiration for FinWell came from a simple observation: people struggle to manage multiple aspects of their lives—finances, investments, and health—often using fragmented tools and platforms. We realized that decision-making in these domains is interconnected. For instance, choosing the right health insurance affects your financial portfolio, while market volatility impacts your ability to afford healthcare.

We envisioned a world where an intelligent, decentralized agent network could seamlessly connect these dots, providing holistic guidance that considers your complete financial and wellness picture. The rise of AI agents and decentralized systems presented the perfect opportunity to build something truly revolutionary.

What We Learned

Building FinWell taught us invaluable lessons across multiple domains:

Technical Mastery:

  • Agent Architecture: We mastered Fetch.ai's uAgents framework, learning how to design autonomous agents that communicate effectively through ChatProtocol
  • Decentralized Communication: Understanding how to deploy agents on AgentVerse and enable cross-domain communication
  • LLM Integration: Working with ASI1 Mini LLM taught us how to optimize AI responses for specific use cases
  • Multi-API Integration: Connecting with stock market APIs, Solana blockchain, and health data sources

Domain Expertise:

  • Financial Analysis: Implementing ratio-based valuation and sentiment analysis for investment decisions
  • Blockchain Technology: Deep diving into Solana's ecosystem for crypto portfolio management
  • Healthcare Intelligence: Building symptom analysis and insurance recommendation systems

System Design:

  • Creating modular, scalable architectures that can handle complex inter-agent communications
  • Designing user experiences that feel natural despite the complexity underneath

How We Built It

Phase 1: Architecture Design We started by mapping out the three core domains (Finance, Crypto, Health) and identifying the specific agents needed for each:

  • Financial Domain: Stock analyst and news sentiment agents
  • Crypto Domain: Solana wallet tracker and token sentiment analyzer
  • Health Domain: Symptom collector, health analyzer, and insurance recommendation agents

Phase 2: Agent Development Each team member took ownership of specific agents:

  • Built individual agents using uAgents framework
  • Implemented ChatProtocol for seamless inter-agent communication
  • Integrated domain-specific APIs and data sources
  • Created the central routing system through advisor_agent.py and main_cli_agent.py

Phase 3: Integration & Orchestration

  • Connected all agents through a unified communication protocol
  • Built the ASI1 wrapper agent to provide intelligent coordination
  • Created the CLI interface for user interaction
  • Deployed agents to AgentVerse for public accessibility

Challenges We Faced

Technical Challenges:

  1. Agent Communication Complexity

    • Challenge: Ensuring reliable communication between 8+ different agents across domains
    • Solution: Implemented robust error handling and fallback mechanisms in ChatProtocol

  2. API Rate Limiting & Data Integration

    • Challenge: Managing multiple external APIs (stock markets, Solana blockchain, health databases) with different rate limits
    • Solution: Built intelligent caching systems and request queuing mechanisms

  3. Real-time Performance

    • Challenge: Users expect instant responses, but complex financial analysis takes time
    • Solution: Implemented asynchronous processing and progressive response delivery

  4. State Management Across Agents

    • Challenge: Maintaining user context across different specialized agents
    • Solution: Created a shared state management system using the advisor agent as a central coordinator

Domain-Specific Challenges:

  1. Health Data Sensitivity

    • Challenge: Handling medical information responsibly while providing useful recommendations
    • Solution: Implemented privacy-first architecture with local processing and anonymized data flows

  2. Financial Data Accuracy

    • Challenge: Ensuring investment recommendations are based on reliable, up-to-date information
    • Solution: Built multi-source verification systems and real-time data validation

What Makes FinWell Special

Innovation in Agent Architecture:

  • First-of-its-kind multi-domain agent network that connects finance, crypto, and health
  • Intelligent routing system that understands user intent across domains
  • Seamless integration with decentralized platforms

User-Centric Design:

  • Natural language interface that makes complex financial and health decisions accessible
  • Personalized recommendations based on complete user profile
  • Real-time insights that adapt to changing market and health conditions

Technical Excellence:

  • Robust, scalable architecture built on proven frameworks
  • Public deployment on AgentVerse for community access
  • Open-source approach encouraging community contributions

Registered Agents on AgentVerse

The following FinWell agents are publicly accessible on AgentVerse for decentralized communication:

Agent Name Role Description
stock_analyst_agent Performs deep equity research and ratio-based valuation analysis
news_summariser Fetches and summarizes sentiment from the latest stock news articles
solana_wallet_agent Fetches real-time wallet balance and transaction data from Solana
solana_token_sentiment_agent Tracks market sentiment and token-level insights for Solana assets
collector_agent Collects user symptoms for medical triage
analyser_agent Evaluates symptoms and prepares health analysis memos
asi1_wrapper_agent Interfaces with ASI1 Mini LLM to provide smart coordination and query resolution
insurance_agent Recommends personalized insurance plans based on user profile and analysis

Future Vision

FinWell represents just the beginning of what's possible when AI agents work together to solve complex, interconnected problems. We envision expanding into additional domains like real estate, education planning, and environmental impact tracking—creating a comprehensive life management ecosystem.

Our journey building FinWell has shown us that the future of personal decision-making lies not in isolated tools, but in intelligent, collaborative agent networks that understand the holistic nature of human needs.

FinWell isn't just a project—it's a glimpse into the future of personalized, intelligent life management.

Built With

Share this project:

Updates