Inspiration
Our ISI AI product focuses on underwriting application ingestion so I wanted to build something for the claims intake workflow.
What it does
A fraud anomaly solution for insurance claims. When a new claim comes in the system scores it against historical claims patterns and flags why it looks risky (or doesn’t) in plain language an adjuster can act on immediately, not just a score.
How we built it
Using Coco to plan out the idea and worked through the design process.
Challenges we ran into
There was no specific data was available to specify fraud so we have to create new data points.
Accomplishments that we're proud of
I never used Snowflake before so it was cool to work with it. I think the solution would be able to be made production ready with just a bit more work.
What we learned
I never used Snowflake before so it was cool to work with it. Coco is a super agent!!
What's next for Trick the Fraudster
Build out the concept and present it to our sales and R&D teams to adopt the project.
Built With
- coco
- python
- snowflake
- streamlit
