TrevorAI

• 

  Portfolio Page

• 

  Transactions Page

Inspiration

Inspired by Trevor Lefkowitz from Ghosts US, TrevorAI is a wealth-management agent designed for everyday investors. Just like the financially savvy ghost Trevor, our AI assistant helps users make informed investment decisions without complex financial jargon. We noticed that many people find the stock market intimidating, and wealth-management services are often accessible only to high-net-worth individuals. TrevorAI bridges this gap by bringing personalized investment assistance to everyone.

What it does

TrevorAI is an AI-powered financial assistant that:

1. Analyzes the user’s portfolio and recommends optimal stocks to buy based on risk profile and goals
   
2. Monitors stock performance in real time and suggests strategic buy/sell timings
   
3. Executes trades automatically when authorized by the user
   
4. Provides a comprehensive dashboard visualizing portfolio performance, stock distributions, and trends
   
5. Maintains a detailed transaction history, distinguishing between AI-initiated and user-initiated trades
   
6. Enables natural phone conversations for stock management—no technical know-how required

How we built it

Our solution combines several technologies:

1. Frontend:
   
   • Framework: Next.js
     
   • Styling: Tailwind CSS
     
   • Visualization: Recharts, Nivo
2. Backend:
   
   • Server: Flask
     
   • Communication: WebSockets & RESTful APIs
3. AI Agent:
   
   • Architecture: Multi-agent system powered by Gemini 2.5 Flash
     
   • Features: Function calling for conversational finance tasks
4. Research Tools:
   
   • Custom integrations for stock research, web search, and market analysis
5. Financial Data:
   
   • APIs: Finnhub, AlphaVantage (real-time stock quotes)
6. Database:
   
   • Storage: Structured JSON for portfolios and transactions
7. Phone Integration:
   
   • Voice tech enabling hands-free interaction with TrevorAI

Challenges we ran into

• Finnhub Module Integration: Adapting the CommonJS finnhub module to work with Next.js server components, handling client vs. server contexts
  
• Real-time Data Rendering: Optimizing performance for complex visualizations with live updates
  
• Natural Language Understanding: Building sophisticated tool-calling logic to parse varied user queries
  
• Market Data Limits: Staying within free-tier API quotas while maintaining accurate portfolio valuations

Accomplishments that we’re proud of

• Built an intuitive dashboard that makes complex financial data accessible
  
• Implemented an AI agent capable of both advising on and executing trades
  
• Created a responsive UI compatible across desktop and mobile
  
• Enabled phone-call interactions to reach users who prefer voice over text
  
• Distinguished AI-initiated vs. user-initiated trades for full transparency

What we learned

• The power of clear, intuitive visualizations for complex datasets
  
• Best practices for integrating AI agents into production apps
  
• Techniques for handling and streaming real-time financial data
  
• Designing cross-platform experiences that blend voice and visual interfaces
  
• Security and responsibility considerations when automating financial transactions

What’s next for TrevorAI

1. Advanced Portfolio Analysis: More sophisticated optimization algorithms
   
2. Expanded Asset Classes: Support for ETFs, bonds, and cryptocurrencies
   
3. Personalized Learning: Adaptive recommendations based on user behavior
   
4. Notifications: Proactive alerts for market events impacting user holdings
   
5. Social Features: Opt-in sharing of portfolios and strategies
   
6. Brokerage Integration: Seamless connectivity with major trading platforms
   
7. Enhanced Research: Deeper trend forecasting and company analysis

Built With

• fastapi
• gemini
• python
• react
• websocket