SLKnight

Inspiration

I built a student loan repayment company back in 2016-19 and received venture funding for the project. One problem I constantly saw was how unique individual outcomes were- there were almost millions of cases which means it was hard to build a one size fits all project. I considered the idea a dead end until with LLMs I saw an opportunity to tailor each individual circumstances to custom strategies to beat the debt trap.

What it does

SLKnight spawns a factory of agents that hatches repayment plans, relocation strategies to make money, automates document scanning and writing with minimal human oversight such that most of the hard parts of navigating the student debt landscape becomes easy to navigate with a few clicks.

How we built it

I found out about this hackathon pretty late, and vibecoded this in Antigravity basically in 3-4 days. I uploaded the rules and requirements from the hackathon along with my own past experience to Gemini chat to come up with project ideas. I used demo data to speed up the process. All I can say its been a whirlwind in the past few days and a lot of Redbulls.

Challenges we ran into

Hitting rate limits on vibe-coding and creating a good interface for AI agents to talk to each other. Since time was short, I decided to forego using financial API like Plaid and real time data like Fed Gov rules and regulations and instead highlight the capacity of the agents to help a borrower run towards financial freedom.

Accomplishments that we're proud of

I like that I solved something that was always at the back of my mind- pretty crazy because new tech can unlock even ideas I thought were dead end.

What we learned

Have a clear idea before building. I originally wanted to build a game that gives out challenges that a student debt borrower must complete, but I realized scope creep was real so I instead decided to take advantage of the agents I have built to showcase their capacity. I think having a clear vision is crucial at the beginning.

What's next for SLKnight

Connecting to data layer via Plaid and Stripe and being able to make payments. iOS app and seeding the product to facebook student loan communities to gain feedback.

Built With

• antigravity
• firebase
• gcp
• gemini
• mapbox
• next.js
• node.js
• tailwind
• typescript