Kairos

Inspiration

Mid-sized companies and and local telecom providers are increasingly vulnerable to scams and document forgeries. We wanted to bridge the gap between AI-powered fraud detection and real-world applicability for these often-overlooked institutions. Our name, Kairos, is inspired by the Greek god of the opportunity moment, which is all about those critical moments and as a way to unite past learnings, present technologies, and future possibilities into one powerful platform. We wanted to spark trust and connection in everyday financial and communication experiences.

What it does

Kairos is a multi-layered AI security system that detects fraud across two major channels. First is with document fraud detection to catch any flags falsified documents like pay stubs, tax returns, or IDs used in loan applications. The next is real time scam call detection to analyze ongoing calls using voice tone, sentiment, and context.

How we built it

Frontend: React, Expo, Node.js. Backend: Flask, Python. Other tools used: Deepgram, Tesseract, Pillow, Gemini, Pdf2Image, Poppler.

Challenges we ran into

Integrating Gemini efficiently on limited compute while maintaining responsiveness. Also had difficulties with devices functionalities which limited our abilities especially when integrating between devices.

Accomplishments that we're proud of

We're proud that we successfully implemented a solution that banks and telecoms could realistically adopt today and evolve with tomorrow. Also various tools we incorporated for both features were new to all of us, so we're proud of successfully integrating them.