FinTerra - DevPost Submission

Inspiration

Traditional financial advisors charge thousands of dollars annually, making expert financial guidance inaccessible to most people. We recognized a stark inequality: high-net-worth individuals get sophisticated portfolio analysis and personalized recommendations, while the average person navigates complex markets alone.

FinTerra was born from a simple belief: everyone deserves institutional-quality financial planning.

We envisioned an AI advisor who knows your complete financial picture, speaks in plain English (not financial jargon), provides actionable recommendations, and is available 24/7. That's how Elowen came to be.

What it does

FinTerra is an intelligent financial management platform that brings expert-level financial planning to everyone through its AI advisor, Elowen.

Unified Dashboard:

  • Track finances with interactive customizable charts
  • Monitor spending across categories in real-time
  • View risk assessment analysis for investment portfolios across multiple accounts
  • Get personalized financial tips based on your behavior

Intelligent Recommendations:

  • AI analyzes your portfolio and suggests optimizations (e.g., "Trim Apple holdings by 40% to reduce risk")
  • Budget analysis following the 50/30/20 rule with specific category adjustments
  • Proactive tips (e.g., "Your spending went over budget by $450 last month")

Monthly Consultation

  • Automated monthly check-ins via a 3-step pipeline: budget analysis and adjustments; portfolio risk and goal progress review; residual-income allocation (savings/investments); next-month strategy

Natural Language Chat:

  • Ask Elowen anything about your finances in plain English
  • Get instant portfolio analysis: "Analyze my investments"
  • Consult on specific concerns: "My credit card utilization is at 35%"
  • Receive concise, jargon-free guidance

Deep Dive Modals:

  • Click any account (checking, savings, credit cards, loans, investments) for detailed transaction history
  • View holdings breakdowns for investment accounts
  • Track progress on goals (debt payoff, emergency fund, etc.)

How we built it

Technology Stack:

  • Frontend: React with Tailwind CSS for responsive, modern UI
  • Backend: FastAPI with Python for robust API design
  • AI Engine: Google ADK for intelligent agent orchestration
  • LLM: Gemini 2.5 Flash for natural language understanding and generation
  • Data Visualization: Recharts for interactive financial charts

Multi-Agent Architecture:

FinTerra uses a sophisticated agent orchestration system where specialized sub-agents handle different research tasks:

Investment Analyzer Agent delegates to:

  • Market Researcher (analyzes sector trends)
  • Diversification Researcher (evaluates asset allocation)
  • Risk Researcher (identifies concentration risks)

Consultant Agent specializes in addressing specific financial tips

Budget Analyzer Agent provides rule-based recommendations

The system uses RAG to enrich AI responses with real-time financial data, ensuring every recommendation is grounded in actual user data.

Challenges we ran into

1. Balancing AI Tone: Getting Elowen to sound like a trusted friend while maintaining financial authority required extensive prompt engineering. We iterated through multiple versions to find the perfect balance between approachable and credible.

2. Data Integration: Unifying financial data across diverse account types (checking, savings, credit cards, loans, investments, 401k) with different schemas required careful modeling and state management.

3. Responsive Formatting: Making AI-generated responses with markdown, bold text, and line breaks digestible for users required extensive testing and custom formatting logic.

Accomplishments that we're proud of

Built a production-ready multi-agent AI system using Google ADK with proper session management and agent orchestration

Created a beautiful, modern UI with dark mode, smooth animations, and responsive design that works flawlessly on all devices

Developed intelligent, context-aware recommendations - our AI actually analyzes portfolios and provides specific, actionable advice (not generic responses) and makes UI changes in real-time

Implemented real-time financial tracking with interactive charts, transaction history, and account deep-dives

Achieved under-250-word responses from our AI that are concise, jargon-free, and immediately actionable

Built a scalable architecture with paginated APIs that can handle thousands of transactions efficiently

Created Elowen's personality - an AI advisor who feels human, remembers context, and talks in plain English

Deployed Application - Deployed a secure application despite continuous deployment challenges with GitHub Actions, DNS related issues, and secure certificate rate limitation throttling

What we learned

Agent Orchestration is Powerful: Breaking complex financial analysis into specialized sub-agents (Market Researcher, Diversification Researcher, Risk Researcher) produces significantly better results than monolithic AI systems. Each agent can focus on what it does best.

UX Matters More Than AI Sophistication: The best AI is useless if users can't understand or act on recommendations. We spent significant time ensuring responses are concise, jargon-free, and immediately actionable.

Financial Apps Demand Trust: Every recommendation must be transparent and explainable. Users need to understand the "why" behind advice, not just the "what."

RAG Changes Everything: Enriching AI responses with actual financial data makes recommendations grounded and specific - not generic financial tips you could get from any blog.

What's next for FinTerra

Enhanced AI Capabilities:

  • Real-time risk engine monitoring portfolio volatility with automated rebalancing recommendations
  • Automated investment actions via stock APIs and payroll integration for seamless money management

Data Integration:

  • Live market data integration for portfolio tracking
  • Credit score monitoring and improvement recommendations

Collaborative Features:

  • Family financial planning with shared dashboards
  • Couples' budget management with synchronization
  • Financial advisor partnership portal

Advanced Analytics:

  • Predictive spending models ("Based on patterns, you'll overspend by $200 this month")
  • Tax optimization recommendations
  • Retirement planning with Monte Carlo simulations

Our Vision: Make FinTerra the default financial planning platform for millions, proving that sophisticated financial advice doesn't need to cost a fortune - it just needs great AI.

Built With

Share this project:

Updates