Inspiration

I recently went through the complex process of getting a loan and had to talk to multiple agents across speciality such as real estate, financing, council regulations, etc. A one stop shop application would have made my job much easier.

What it does

The app gets loan contracts as input, analyzes using LLM to detect key clauses and translates them in simple language. This is done for free. For detailed analysis, the user can pay a nominal fee using MNEE stablecoin. There are two other modules for highlighting red flags in the contract as well as providing negotiation strategies that the borrower can use against the lender to get favorable terms.

How we built it

Itertively going from a simple prototype using langchain components that can analyse document and integrating crypto payments to test its functionality. The initial tests were done using sepolia testnet payments before transitioning to real stablecoin payments using eth. The prototype was then converted into a real productionalizable product deployed on google cloud. Finally its about incorporating other modules on red flags and negotiation.

Challenges we ran into

Payments integration was complex due to the unavailability of the required contracts for stablecoin. The scaling up for modules had some challenges before I used a modular architecture to separate out the agents and ensure they work separately and accurately.

Accomplishments that we're proud of

Developing a production grade product at the intersection of AI and Stablecoin for a trillion dollar loan market is an accomplishment I'm proud. It's the first step to more complex products.

What's next for Loan Mate

Develop multi agent architecture which acts like a moot court challenging the borrower by role playing the lender to help them practice for negotiations.

Value proposition

Loan Mate turns opaque loan contracts into transparent, negotiable financial products.

Scalability

Every new customer costs almost nothing The same system works for: Mortgages Auto loans Student loans Business financing

Efficiency

This compresses an entire professional services stack into software:

Before: Borrower → Broker → Lawyer → Bank → Council → Agent → Delay → Fees → Mistakes

After: Borrower → Upload → AI → Actionable guidance → Pay

Built With

• google-cloud
• langchain
• python
• typescript