Financial Wellness Agent (FWA)

AI Chat
AI Powered Saving Goals
Financial Reports
Dashboard
AI Market Analysis
Market View
AI Recommendation

General info

The Financial Wellness Agent (FWA) is a smart, AI-driven personal finance companion designed to make managing money as easy as having a chat. It leverages Google's Gemini models to understand natural language, allowing users to log transactions, analyze spending, and get personalized financial advice simply by talking or texting.

Inspiration

Tracking finances is often tedious. Manual entry apps feel like data entry jobs, and automated bank trackers often lack context. I wanted to build a bridge between the two—a system that understands "I just spent 50k on lunch" without making me fill out a form. The goal was to create a financial assistant that feels human, proactive, and genuinely helpful, empowering users to improve their financial health with minimal effort.

What it does

FWA is a mobile-first application that serves as your proactive financial partner.

Natural Language Transaction Logging: Simply types or say "Spent 50k on warmindo" or "Salary came in, 10 million", and FWA intelligently categorizes and logs it.
Intelligent Insights: Ask questions like "Can I afford a new phone?" or "How much did I spend on coffee this month?" and get data-backed answers.
Spending Analysis: Detects patterns, anomalies, and opportunities to save.
Goal Tracking: Helps you set realistic savings goals and tracks your progress with specific advice on how to reach them faster.
Receipt Scanning: Snap a photo of your receipt, and FWA uses Gemini 3 Multimodal capabilities to instantly extract merchant, items, and total amount—no typing required.
Voice Interaction: Talk to your agent directly for a hands-free experience.

How I built it

The project is built on a modern, scalable stack:

AI Core: Google Gemini (Gemini 3 Flash / Gemini 3 Pro) drive the core logic. Gemini 3 Flash handles high-speed transaction parsing and chat, while Gemini 3 Pro provides deep reasoning for financial analysis and goal feasibility assessments.
Frontend: Built with Next.js 14 for a responsive, app-like experience, styled with TailwindCSS and shadcn/ui. Framer Motion adds fluid animations to make finance feel less rigid.
Backend: FastAPI (Python) serves as the brain, orchestrating the agents.
Database: PostgreSQL (via Supabase/Neon) for structured data and Redis for high-speed caching.
Agentic Architecture: Inspired by advanced multi-agent patterns, the system uses specialized sub-agents (Planner, Transaction Parser, Budget Advisor) to handle complex tasks reliably.

Challenges I ran into

Structured Data Extraction: Getting an LLM to consistently output strict JSON for database transactions from messy human language was tricky. I used Gemini's structured output capabilities and validation layers to ensure 100% reliability.
Latency: Making the chat feel real-time while performing complex analysis. optimizing the agent pipeline and using Gemini Flash models helped significantly.
Context Management: Ensuring the AI remembers your "current balance" and "savings goals" seamlessly across the conversation required a robust context injection strategy.

Accomplishments I'm proud of

Seamless Parsing: The ability to type "50k lunch" and have it instantly logged is a game-changer for usability.
Multi-Agent Orchestration: Successfully implementing a system where different "specialists" (Planner, Reporter, Advisor) collaborate to solve user requests.
UI/UX: Creating a financial dashboard that doesn't look boring—it looks and feels like a modern consumer app.

What I learned

I deepened my understanding of Agentic AI workflows, specifically how to decompose complex problems (financial planning) into atomic tasks for different AI agents. I also learned the power of Google's Gemini models in handling multimodal and long-context tasks, which is crucial for analyzing months of financial history.

What's next for Financial Wellness Agent

Deeper Integrations: Connecting directly to bank APIs for automated reconciliation.
Investment Agent: Adding a specialized agent for investment portfolio tracking and advice.
Gamification: Adding challenges and badges to make saving money fun.