Making financial planning as accessible as checking WhatsApp.
Turning confused savers into confident investors — in seconds, at zero cost.
📖 Docs • 🚀 Quick Start • 🏗️ Architecture • 🤝 Contributing
95% of Indians have no financial plan.
Traditional financial advisors charge ₹25,000+ per year and cater exclusively to High Net-worth Individuals (HNIs). Hundreds of millions of middle-class Indians navigate SIPs, tax regimes, insurance gaps, and retirement planning completely alone.
Concierge democratizes expert-grade financial advice using AI.
| Feature | Description |
|---|---|
| 🔥 FIRE Path Planner | 4-step wizard generating personalized retirement roadmaps |
| 🤖 AI Mentor Chat | LLM-powered conversational advisor with RAG |
| 👤 Profile Dashboard | Track FIRE progress, corpus targets, SIP metrics |
| 🌐 Multilingual | English, Hindi & Hinglish support |
| 🔒 Secure & Private | Rate limiting, input validation, session isolation |
The core feature guides users through a 4-step onboarding wizard and outputs a complete financial roadmap.
flowchart TD
A([👤 User Starts]) --> B
subgraph WIZARD ["📋 4-Step Onboarding Wizard"]
B[Step 1\nName & Current Age]
B --> C[Step 2\nIncome · Expenses · Savings · Investments]
C --> D[Step 3\nTarget Retirement Age\nPost-Retirement Expenses]
D --> E[Step 4\nRisk Profile · Language Preference]
end
E --> F{⚙️ FIRE Calculator Engine}
subgraph ENGINE ["🧮 calculator.py"]
F --> G[Inflation Adjustment\n6% annual rate]
G --> H[Corpus Calculation\n25x–35x Rule + 10% healthcare buffer]
H --> I[Risk-Based Returns\n8% / 11% / 14% CAGR]
I --> J[Existing Wealth Projection\nCompound forward & deduct]
J --> K[SIP Calculation\nPMT formula on corpus gap]
K --> L[FIRE Classification\nLean / Moderate / Fat]
end
L --> M{🤖 AI Roadmap Generator}
subgraph AI ["🧠 ai_advisor.py — LLM + RAG"]
M --> N[Retrieve Financial Knowledge\nFAISS Vector Store]
N --> O[Generate Personalized Plan\nDeepSeek via LangChain]
end
O --> P
subgraph OUTPUT ["📊 Results Dashboard"]
P[FIRE Corpus Range\nMin–Max]
P --> Q[Monthly SIP Range]
Q --> R[Years to FIRE + FIRE Type]
R --> S[Asset Allocation by Risk Profile]
S --> T[6-Month Action Plan\nMonthly Milestones]
T --> U[Tax Optimization\n80C · NPS · ELSS]
U --> V[Emergency Fund Target]
V --> W[Insurance Recommendations\nTerm + Health]
end
style WIZARD fill:#1a1a2e,stroke:#e94560,color:#fff
style ENGINE fill:#16213e,stroke:#0f3460,color:#fff
style AI fill:#0f3460,stroke:#533483,color:#fff
style OUTPUT fill:#1a1a2e,stroke:#2ecc71,color:#fff
flowchart LR
subgraph FRONTEND ["🖥️ Frontend — Vanilla HTML/CSS/JS + Tailwind"]
FP1[firststep_page.html\nStep 1]
FP2[fire_wizard.html\nStep 2]
FP3[thirdstep.html\nStep 3]
FP4[fourthstep.html\nStep 4]
FPR[fire_plan.html\nResults]
FPC[ai_mentor.html\nChat]
FPRO[profile_page.html\nProfile]
end
subgraph BACKEND ["⚙️ Backend — Python FastAPI"]
MAIN[main.py\nRoutes · Middleware · CORS]
MODEL[model.py\nPydantic v2 Validation]
CALC[calculator.py\nFIRE Math Engine]
AI[ai_advisor.py\nLLM + RAG Layer]
DB[database.py\nSupabase / In-Memory]
end
subgraph INFRA ["☁️ Infrastructure & AI"]
SUPA[(Supabase\nPostgreSQL)]
FAISS[(FAISS\nVector Store)]
DS[DeepSeek LLM\nvia LangChain]
GEMINI[Gemini\nIntegration]
ST[Sentence Transformers\nEmbeddings]
end
FRONTEND -->|REST API Calls| MAIN
MAIN --> MODEL
MODEL --> CALC
MODEL --> AI
CALC --> DB
AI --> FAISS
AI --> DS
AI --> GEMINI
FAISS --> ST
DB --> SUPA
style FRONTEND fill:#1a1a2e,stroke:#e94560,color:#fff
style BACKEND fill:#16213e,stroke:#0f3460,color:#fff
style INFRA fill:#0f3460,stroke:#533483,color:#fff
sequenceDiagram
participant U as 👤 User
participant FE as 🖥️ Frontend
participant API as ⚙️ FastAPI
participant RAG as 🔍 RAG (FAISS)
participant LLM as 🧠 DeepSeek LLM
participant DB as 🗄️ Supabase
U->>FE: Sends question (EN/HI/Hinglish)
FE->>API: POST /chat/message
API->>DB: Fetch session + FIRE plan context
DB-->>API: User profile & plan data
API->>RAG: Retrieve relevant financial docs
RAG-->>API: Top-k chunks
API->>LLM: Prompt = question + context + RAG chunks
LLM-->>API: Personalized response
API->>DB: Save message to history
API-->>FE: Response text
FE-->>U: Display answer
flowchart TD
A[Monthly Post-Retirement Expenses\ne.g. ₹60,000] --> B
B["Inflation Adjust to Retirement Year\nExpenses × (1.06)^years_to_retire"]
B --> C
C["Calculate Annual Expenses\nMonthly × 12"]
C --> D
D["FIRE Corpus Range\nMin = 25 × Annual Expenses\nMax = 35 × Annual Expenses\n+ 10% Healthcare Buffer"]
D --> E
E["Project Existing Wealth\nSavings + Investments × (1 + CAGR)^years"]
subgraph CAGR_TABLE ["📊 Risk-Based CAGR"]
C1[🛡️ Conservative\n8% CAGR]
C2[⚖️ Moderate\n11% CAGR]
C3[🚀 Aggressive\n14% CAGR]
end
E --> F["Remaining Corpus Gap\nTarget Corpus − Projected Wealth"]
F --> G["Monthly SIP\nPMT formula on remaining gap"]
G --> H{FIRE Type Classification}
H -->|< ₹30k/month| I[🟡 Lean FIRE]
H -->|₹30k–₹1L/month| J[🟢 Moderate FIRE]
H -->|> ₹1L/month| K[🔵 Fat FIRE]
style CAGR_TABLE fill:#16213e,stroke:#0f3460,color:#fff
concierge/
│
├── 📂 backend/
│ ├── 🐍 main.py # FastAPI app, routes, middleware, CORS
│ ├── 🐍 model.py # Pydantic v2 input models & injection protection
│ ├── 🐍 calculator.py # FIRE math engine (corpus, SIP, allocation)
│ ├── 🐍 ai_advisor.py # AI roadmap generation via LLM + RAG
│ ├── 🐍 database.py # Supabase + in-memory fallback layer
│ └── 📄 requirements.txt
│
└── 📂 frontend/
├── 🌐 firststep_page.html # Step 1 — Name & Age
├── 🌐 fire_wizard.html # Step 2 — Income & Savings
├── 🌐 thirdstep.html # Step 3 — FIRE Goals
├── 🌐 fourthstep.html # Step 4 — Risk & Language
├── 🌐 fire_plan.html # Results dashboard
├── 🌐 ai_mentor.html # Chat interface
├── 🌐 profile_page.html # User profile
├── 🎨 style.css
└── ⚡ app.js
- Python 3.10+
- Node.js (optional — for frontend tooling only)
# 1. Clone the repository
git clone https://github.com/your-org/concierge.git
cd concierge
# 2. Install Python dependencies
pip install -r backend/requirements.txtCreate a .env file inside the backend/ directory:
# ─── Supabase (optional — falls back to in-memory if not set) ───────────────
SUPABASE_URL=your_supabase_project_url
SUPABASE_KEY=your_supabase_service_role_key # Use service_role, NOT publishable key
# ─── LLM ────────────────────────────────────────────────────────────────────
DEEPSEEK_API_KEY=your_deepseek_api_key
# ─── CORS ───────────────────────────────────────────────────────────────────
ALLOWED_ORIGINS=http://localhost:8000,http://127.0.0.1:8000💡 Note: The app runs fully without Supabase using an in-memory fallback. Data will not persist across server restarts in this mode.
cd backend
uvicorn main:app --reload --port 8000Open your browser at http://127.0.0.1:8000 🎉
| Method | Endpoint | Description |
|---|---|---|
POST |
/fire-plan |
Generate a complete FIRE plan from user inputs |
POST |
/chat/start |
Start a new AI Mentor chat session |
POST |
/chat/message |
Send a message to the AI Mentor |
GET |
/chat/history/{session_id} |
Retrieve full chat history |
GET |
/chat/sessions/{user_id} |
List all sessions for a user |
DELETE |
/user/{user_id} |
Delete all user data (GDPR compliant) |
GET |
/search?query=... |
Search the RAG knowledge base |
curl -X POST http://127.0.0.1:8000/fire-plan \
-H "Content-Type: application/json" \
-d '{
"name": "Arjun Mehta",
"age": 30,
"monthly_income": 150000,
"monthly_expenses": 70000,
"current_savings": 500000,
"existing_investments": 200000,
"fire_target_age": 50,
"monthly_expenses_post_fire": 60000,
"risk_profile": "moderate",
"language": "english"
}'📦 Sample Response (click to expand)
{
"fire_corpus": {
"min": 28500000,
"max": 39900000
},
"monthly_sip": {
"min": 42000,
"max": 58000
},
"years_to_fire": 20,
"fire_type": "Moderate FIRE",
"asset_allocation": {
"equity": "60%",
"debt": "30%",
"gold": "10%"
},
"emergency_fund_target": 420000,
"action_plan": ["Month 1: ...", "Month 2: ..."],
"tax_suggestions": ["80C: ₹1.5L via ELSS", "NPS: Additional ₹50K deduction"],
"insurance": {
"term": "₹1 Cr cover recommended",
"health": "₹10L family floater"
}
}flowchart LR
A[Incoming Request] --> B{Body Size Check\n< 1 MB}
B -->|Pass| C{CORS Validation\nAllowed Origins Only}
C -->|Pass| D{Rate Limiter\nSlowAPI}
D -->|Pass| E{Pydantic v2\nInput Validation}
E -->|Pass| F{Injection Scanner\nPattern Matching}
F -->|Pass| G[✅ Process Request]
B -->|Fail| X1[❌ 413 Payload Too Large]
C -->|Fail| X2[❌ 403 Forbidden]
D -->|Fail| X3[❌ 429 Too Many Requests]
E -->|Fail| X4[❌ 422 Validation Error]
F -->|Fail| X5[❌ 400 Bad Request]
style G fill:#2ecc71,color:#000
style X1 fill:#e74c3c,color:#fff
style X2 fill:#e74c3c,color:#fff
style X3 fill:#e74c3c,color:#fff
style X4 fill:#e74c3c,color:#fff
style X5 fill:#e74c3c,color:#fff
| Layer | Measure |
|---|---|
| 🚦 Rate Limiting | 5 req/min /fire-plan · 20 req/min chat · 200 req/day global |
| 📦 Body Size Cap | 1 MB limit via middleware |
| 🔍 Injection Guard | Inputs scanned against known prompt injection patterns |
| 🌐 CORS | Only explicitly allowed origins accepted |
| ✅ Input Validation | Pydantic v2 strict types, ranges & length constraints |
| 🔒 Session Isolation | Chat summaries keyed by session ID — never leaked across users |
| Language | Code | Description |
|---|---|---|
| 🇬🇧 English | english |
Full professional financial English |
| 🇮🇳 Hindi | hindi |
Complete Hindi with financial terms preserved (SIP, ELSS, PPF) |
| 🗣️ Hinglish | hinglish |
Natural Hindi + English mix, as spoken by urban Indians |
Language preference flows through both the AI roadmap generator and the chat advisor for a fully consistent experience.
gantt
title Concierge — Feature Roadmap
dateFormat YYYY-Q[Q]
axisFormat %Y Q%q
section ✅ Shipped
FIRE Path Planner :done, 2025-01-01, 90d
AI Mentor Chat :done, 2025-01-01, 90d
Multilingual Support :done, 2025-01-01, 90d
section 🚧 Coming Soon
Money Health Score :active, 2025-04-01, 60d
Life Event Advisor :2025-05-01, 60d
Tax Wizard (Form 16) :2025-06-01, 60d
section 🔮 Future
Couple's Money Planner :2025-08-01, 90d
MF Portfolio X-Ray :2025-10-01, 90d
| # | Feature | Status |
|---|---|---|
| 💰 | Money Health Score — 6-dimension financial wellness assessment | 🚧 In Progress |
| 💍 | Life Event Advisor — Bonus, inheritance, marriage, new baby planning | 📋 Planned |
| 📄 | Tax Wizard — Form 16 upload + old vs. new regime comparison | 📋 Planned |
| 👫 | Couple's Money Planner — Joint income optimization across both partners | 🔮 Future |
| 📊 | MF Portfolio X-Ray — CAMS/KFintech statement upload with XIRR & overlap analysis | 🔮 Future |
Contributions are welcome! Here's how to get involved:
flowchart LR
A[🍴 Fork the Repo] --> B[🌿 Create Feature Branch\ngit checkout -b feature/your-feature]
B --> C[💻 Make Changes]
C --> D[✅ Commit\ngit commit -m 'Add feature']
D --> E[📤 Push\ngit push origin feature/your-feature]
E --> F[🔃 Open Pull Request]
F --> G[🎉 Merged!]
- Fork the repository
- Create your feature branch:
git checkout -b feature/money-health-score - Commit your changes:
git commit -m 'Add Money Health Score module' - Push to the branch:
git push origin feature/money-health-score - Open a Pull Request
For major changes, please open an issue first to discuss what you'd like to change.
Built with ❤️ by two engineers from RGIPT for the Economic Times AI Hackathon.
 Apurva Sinha @apurvafx🧠 Backend · RAG Pipeline · FIRE Engine · Database · Security Electrical Engineering @ RGIPT · AI/ML Developer Deep Learning · Computer Vision · Embedded Systems 
|
 Aashish Chandra @Aashish-Chandr🎨 UI/UX Design · Frontend Development · Backend Integration Undergrad @ RGIPT · Computer Vision Enthusiast Vehicle Plate Detection · Emotion Recognition · Deep Learning 
|
 Prateek Raj @prs-24💻 Frontend Development · Database Electrical Engineering @ RGIPT · AI/ML Developer Deep Learning · Computer Vision · Embedded Systems · IoT 
|
This project is licensed under the MIT License — see the LICENSE file for details.
