LendingEdge AI

• 

  this is how users interuct with the agentic work flow

Inspiration

My inspiration stems from my firsthand experience as a Loan Officer at a financial institution. I witnessed the devastating impact of high Non-Performing Loan (NPL) ratios, which threatened the institution's stability and limited credit access for honest borrowers. The "blind spot" was clear: we were making lending decisions based on static, outdated paper trails that failed to capture the real-time financial velocity of modern business. I realized that if we could harness the rich data within mobile money ecosystems like M-Pesa and combine it with tailored business advisory, we could transform "high-risk" borrowers into sustainable success stories.

What it does

IMPORTANT:to test the application with an already registered details use test_user@gmail.com as email and test1234 as password , after filling and submitting the form scroll down for the downloadable pdf advice . LendingEdge AI is a comprehensive credit intelligence and business growth platform. It allows borrowers to authorize the secure retrieval of their mobile money transaction data through an Agent-to-Agent interaction protocol. The application analyzes this data to provide a real-time "Financial Health" score. Beyond just a score, the tool asks for a business description and operation level (Startup vs. Growing) to provide a customized roadmap. It advises the user on the optimal loan amount, provides a domain-specific investment strategy, and highlights key risk areas. It also features a social success network where users can connect and learn from peer success stories to foster a culture of repayment and growth.

How we built it

The application is built with a focus on commercial scalability and data integrity. We developed a robust data processing engine to parse transaction histories into actionable liquidity and debt-service metrics. The logic layer uses a risk-assessment framework that calculates: $$Risk\ Index = \frac{Total\ Debt\ Obligations}{\text{Average Monthly Revenue}} \times \sigma_{cashflow}$$ We integrated a generative advisory module that maps business descriptions to industry-specific risk profiles. For the frontend, we designed a web-based interface that prioritizes clarity for both the lender and the borrower, ensuring that complex financial analytics are presented as simple, actionable insights.

Challenges we ran into

One of the primary challenges was designing the Agent-to-Agent interaction protocol to ensure it met strict data privacy standards while remaining user-friendly. Translating raw, messy transaction data into a standardized "Financial Health" score required significant fine-tuning of our analysis algorithms. Additionally, creating a "context-aware" advisory system that could distinguish between the cash flow needs of a retail shop versus a service-based startup required deep integration of business domain logic.

Accomplishments that I am proud of

I am particularly proud of moving beyond a simple "Yes/No" lending tool to a "How-To" growth platform. Successfully automating the extraction of deep financial insights from mobile money data which is often the only financial record for millions of SMEs is a major milestone. I am also proud of the Social Success Network integration, which adds a human element to the digital loan market, encouraging communal accountability.

What we learned

I learned that data without context is a liability, but data with advisory is an asset. My time as a loan officer taught me that many defaults happen not because of bad intent, but because of poor capital deployment. Through this build, I learned how to bridge the gap between "Big Data" and "Small Business" needs. We also gained deep insights into the importance of Zero-Knowledge authorization in building trust within the Fintech ecosystem.

What's next for LendingEdge AI

The next step is to move from CSV-based prototyping to live API integrations with major mobile money providers across the EMEA region. We plan to enhance our predictive model to include seasonal trend forecasting for agricultural businesses. Ultimately, we aim to partner with LMA-affiliated financial institutions to integrate LendingEdge AI into their core banking systems, turning it into a global standard for reducing NPLs in the multi-trillion dollar loan market.

Built With

• css
• flask
• gemini
• html5
• python
• sqlalchemy