Eagle Eyes

A self-improving AML/CFT agent for financial institutions and auditors.

Comment

Inspiration

I work at a currency exchange company and have seen auditors from the central bank of Kuwait come and take 2-3 months auditing our transactions which would take lots of effort and yet not enough accuracy.

What it does

It automates the auditory process, identifying violating transactions based on the Anti-money Laundering (AML) and Counter-terrorist financing (CFT) rules.

How we built it

I used gemini-flash-3.5 with google's antigravity for developing the project along with Arize and its Phoenix platform for the AI observability/traces and MongoDB's Atlas for the database.

Challenges we ran into

Honestly, this was my first hackathon and everything was challenging:) .From gathering all the related rules to setting up the APIs and MCP servers.

Accomplishments that we're proud of

I'm proud of building this project as a whole, since it's my first agent and hackathon participation. As a self-taught developer, this is a huge milestone for me. Learning about AI traces and building the self-improvement loop(Which I never thought that I would pull it off at this stage) is also a great moment for me. And lastly, this was a wild roller-coaster. I never that I would finish the project tbh. Because the problems just wouldn't disappear. But in the end, they did :)!

What we learned

AI traces and observability - resilience - handling anxiety - antigravity's platform

What's next for Eagle Eyes

It could become the next auditor/compliance officer in financial institutions. It just needs a bit of investment.

Built With

Share this project:

Updates