LoanGenie

• 

  Our way of solving

Inspiration

We were inspired by how slow, manual, and frustrating loan processing still is. Customers face delays, repeated verifications, and confusing steps. We wanted to build a faster, smarter, AI-driven solution.

What it does

LoanGenie is an AI-powered personal loan assistant that uses multiple autonomous agents to handle the entire loan journey — inquiry, KYC verification, document checks, credit scoring, eligibility, and sanction letter generation.

How we built it

We used React, FastAPI, MongoDB, Pinecone, and agent frameworks like LangGraph and CrewAI. Each agent (Sales, Verification, Underwriting, Sanction) communicates through structured JSON while the Master Agent orchestrates the workflow.

Challenges we ran into

Coordinating multiple agents, ensuring deterministic responses, handling memory retrieval, designing mock KYC/CRM APIs, and creating a smooth conversational interface were major challenges.

Accomplishments that we're proud of

Building a fully functional multi-agent loan system that feels like a digital loan department and completing an end-to-end automated approval flow.

What we learned

Agentic AI design, context management, financial workflow simulation, and creating realistic document verification pipelines.

What's next for LoanGenie

Voice-based agents, real API integration, multilingual support, and expanding to other financial products.

Built With

• crewai
• fastapi
• javascript
• langgraph
• mongodb
• pinecone
• python
• pytorch
• react
• scikit-learn
• tensorflow