Credigenie

Inspiration

What it does

Inspiration

Loan applications are still slow, manual, and frustrating for both customers and banks. We had a simple question that inspired us: Why can’t loan approval feel as smooth as chatting with a human advisor? CrediGenie was created to rethink loan journeys using agentic AI. With CrediGenie, conversations don’t just happen—they get the job done.

What it does

CrediGenie is an agentic AI loan assistant that conducts persuasive, human-like loan conversations through a web chatbot. It verifies KYC, evaluates credit eligibility, requests documents when needed, and generates a downloadable PDF sanction letter—all from start to finish in minutes.

How we built it

We designed a Master Agent as the conversational manager, coordinating several Worker Agents—Sales, Verification, Underwriting, Document Validation, and Sanction Letter Generator. The system connects with mock banking APIs (CRM, Offer Mart, Credit Bureau), applies clear underwriting rules, and calculates EMI using the formula:

EMI = P * r * (1 + r)^n / ((1 + r)^n - 1)

It delivers results through a user-friendly chat interface that allows uploads and PDF downloads.

Challenges we ran into

We faced challenges in designing a multi-agent orchestration flow while maintaining conversation continuity. We had to handle edge cases like low credit scores, missing documents, and API failures smoothly. Balancing a persuasive sales tone with ethical, non-spammy AI behavior was also crucial. We needed to make complex underwriting logic understandable to non-technical users.

Accomplishments that we're proud of

We completed the entire loan journey within a chatbot. We achieved deterministic underwriting with clear, explainable decisions. Our system generated realistic PDF sanction letters automatically. We handled edge cases well, providing fallback paths. We created a commercially viable, desktop-first prototype that fits into real banking workflows.

What we learned

We found that agentic AI is most effective when roles are clearly defined. Orchestration is more important than the size of the model, and trust comes from transparency. Explaining why a loan was approved or rejected is just as vital as the decision itself.

What's next for CrediGenie

We plan to integrate real credit bureau and KYC systems. We aim to offer multilingual and voice-based loan conversations. We will work on ML-driven conversion optimization and offer personalization. We also intend to implement secure digital signatures and audit logs for production use.

CrediGenie is not just a chatbot—it’s a loan officer that actually gets things done.

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for Credigenie

Built With

• and
• and-llm-(watson-assistant-or-openai-compatible-model)-for-master-agent-orchestration.-microservice-based-worker-agents
• and-reportlab-for-pdf-generation.-git-and-github-for-version-control
• api
• credit-bureau)
• docker-for-local-container-deployment
• fastapi
• flask
• for
• javascript
• json-based-data-storage
• mock-rest-apis-(crm
• offer-mart
• postman
• python
• react
• streamlit