FinGoal AI

# **Project Brief: FinGoal AI**

## Executive Summary
Initial competitive advantage: **AI-first** model + "single advisor" experience that integrates multiple agents, focusing on personalization through real user data (income, expenses, debt, assets, goals), while partnering with financial partners to provide implementation products (opening savings accounts, buying funds, investment referrals). The goal is to help users build financial discipline, save, invest according to their capabilities, and progress toward their financial goals.

The core innovation lies in its multi-agent AI system orchestrated through n8n workflows, where specialized agents handle distinct tasks ensuring greater efficiency, accuracy, and context-aware responses through a persistent memory mechanism.

## Problem Statement
Vietnamese individuals and families face significant challenges in personal financial management due to multiple systemic issues:

1. **Low Financial Literacy** — Many people don't understand financial options, lack long-term planning, and make emotional investment decisions.
2. **Product-Focused Advisory Services** — Current services (banks/insurance/securities companies) lack independent, personalized advice and focus primarily on selling products.
3. **High Traditional Advisory Costs** — Professional financial advice is expensive and unsuitable for low-income individuals or newly affluent F1-F2 groups.
4. **Lack of Discipline & Tracking** — Users need reminders, alerts, and actionable roadmaps they can actually follow.
5. **Insufficient Products for New Individual Investors** — Limited micro-investing, simple robo-advice, and interactive education options.

These barriers lead to missed financial opportunities, unmanaged debt, emotional decision-making, and a reactive approach to personal finance management.

## Market Opportunity & Evidence

The Vietnamese market presents compelling evidence for a personal financial advisory solution:

* **Rapid Growth in Individual Investors**: By end of 2024, there are ~9.2-9.3 million securities accounts (equivalent to ~9% of population), showing a massive base of people interested in personal investment.
* **High Internet & Mobile Penetration**: Internet penetration ~79% (2024) and continuing to grow → suitable for digital financial products.
* **Expanding Middle Class & Rising Income**: Middle-class rapidly expanding (projected to reach ~26% of population by 2026) — more people will have wealth management needs.
* **Fintech & Micro-investing Investment Traction**: Examples like Finhay raising $25M Series B show investor confidence in Vietnam's micro-wealth/fintech market.

**Conclusion**: The market has a broad user base + capability to transition to digital services; there's a large "pie" for personal financial advisory products, especially if they're affordable, accessible, and practical.

## Proposed Solution

### Value Propositions
* **Personalized Advice from Day One** — Creates Personal Financial Plans (PFP) based on reality: income, goals, debt, risk tolerance, and timeframe.
* **Low Cost / Freemium Model** — Anyone can use the basic version for free; paid plans for in-depth planning or human advisor consultations.
* **Actionable Recommendations** — Not just suggestions but **implementation steps**, reminders, and links to open appropriate accounts/funds/products.
* **Integrated Financial Education** — Micro-lessons, behavioral nudges, and gamification to improve financial habits.
* **Security & Transparency** — Private data protection, security standards, and transparent monetization approach.

### Core Features (MVP → v1)

**Smart Onboarding**
* Basic data collection: age, income, average expenses, goals (house purchase, emergency fund), debt, current investments, insurance.
* Optional automatic connection: read bank data (open banking/screen scraping with user permission) or upload images/receipts.

**Core Engine — "Personal Advisor"**
* **Situation Analysis**: Health score, liquidity ratio, expense categories, cash runway.
* **Planning**: Weekly/monthly budgets, savings roadmap, basic insurance recommendations, sample investment portfolios by risk profile.
* **Monitoring & Alerts**: Budget deviations, delayed goals, increasing debt risks.
* **Action/Implementation**: Partner links to open savings accounts, funds, brokerages, or purchase micro-insurance.
* **Personalized Education**: Micro-lessons based on user errors/capabilities.

**UX: Single Virtual "Advisor"**
Users chat with one avatar/advisor; behind the scenes, multiple agents coordinate. Experience must be smooth: 5-7 step onboarding, simple dashboard, actionable checklist.

## Target Users
**Primary User Segment: The Financially Fragmented Professional**
*   **Profile:** Vietnamese professionals and small business owners aged 25-45 with a middle-to-high income.
*   **Pain Points:** They need a single picture of their finances, desire simple guidance, want to be more proactive, and need timely reminders.
*   **Goals:** To achieve major life goals (home, education), reduce debt, build a sustainable budget, and make smart investments.

## Goals & Success Metrics (Hackathon Focus)
*   **Project Objectives:** Demonstrate a functional MVP, validate the multi-agent AI concept, and showcase the technical innovation of the Flutter/n8n/PostgreSQL stack.
*   **User Experience Success Criteria:** Effortless onboarding via a simple form, immediate value through a personalized summary, and an engaging chat interface that accepts both text and voice input.

## MVP Scope (Implemented)
*   **Core Features Delivered:** 
    - Complete email/password authentication system with JWT tokens
    - Multi-agent AI system with 8 specialized workflows (Intent/Session, Orchestrator, Data Collector, Consultant, Plan Maker, Change Adder, Educator, Memory Updater)
    - Unified chat interface with retry handling and error recovery
    - Comprehensive financial dashboard with accounts, budgets, and transactions
    - Full user profile management and spending categories
    - Bilingual support (English/Vietnamese)
*   **Technical Implementation:** PostgreSQL database with UUID v7, n8n workflow orchestration, Flutter Material 3 UI, Provider state management
*   **MVP Success Criteria Met:** Users can register/login, interact with specialized AI financial advisors, view comprehensive financial dashboard, manage profiles and categories, all within a polished mobile interface.

## Post-MVP Vision
Building upon the robust MVP foundation, future enhancements include:
- **Advanced Analytics**: Enhanced trend analysis, predictive modeling, and personalized financial insights
- **Bank Integration**: Direct API connections to Vietnamese banks and financial institutions
- **Real-time Features**: Socket.io integration for live notifications and collaborative financial planning
- **Premium AI Agents**: Specialized agents for investment analysis, insurance optimization, and tax planning
- **Social Features**: Family financial planning, shared budgets, and financial goal collaboration

## Technical Architecture

**Hackathon Goal**: Minimize complexity for quick demo, easy deployment and modification.

* **Frontend**: Flutter (mobile) — single codebase for iOS/Android (web version can be added later).
* **Orchestration/Backend**: n8n as API gateway + workflow/AI orchestration (connecting LLM, data transfer, third-party calls). Add small microservices (Node.js) if complex logic needed.
* **AI**: Call LLM through n8n workflows; basic RAG with lightweight vector store (local or managed) to answer based on policy/user data.
* **Database**: PostgreSQL with UUID v7, JWT authentication.
* **Integrations**: Design connectors for open banking/brokerage APIs but use mock/sandbox during hackathon phase.
* **Deployment**: Cloudflare (Pages/Workers + CDN) for frontend and edge proxy; containerized backend (Docker) or serverless, quick cloud deployment when needed.
* **Security (planned)**: OAuth2/JWT, TLS, encrypt at rest, secret management (KMS). Hackathon prioritizes secure-by-design but implements hardening per backlog.

## Constraints & Assumptions
*   **Constraints:** Hackathon timeline, defined tech stack (Flutter, n8n, PostgreSQL), no live financial data.
*   **Assumptions:** N8N can handle the required complexity with acceptable performance; a Flutter package can render forms from JSON; speech-to-text is sufficiently accurate; OCR can be simulated.

## Challenges & Risk Mitigation

1. **Trust & Credibility**: Risk that users won't trust AI to manage their money.
   * *Mitigation*: Algorithm transparency, explainability, "talk to human" option, public case study trials.

2. **Data & Security**: Risk of data leaks/insufficient security.
   * *Mitigation*: High security standards, SOC2-type controls, encryption, auditing.

3. **AI Errors/Hallucination**: AI providing incorrect financial recommendations.
   * *Mitigation*: Human-in-the-loop for critical financial advice; validation layer; logging & rollback capabilities.

4. **Technical Risks**: N8N performance latency, complexity of dynamic UI rendering in Flutter, prompt engineering for structured JSON output, and hackathon scope creep.