TrustBill

Inspiration

Invoice fraud is one of the most pervasive financial crimes faced by enterprises today, costing businesses billions globally every year. Recently I came across this article which inspired talked about how people can prevent invoice fraud by manually checking them before proceding. This let me to think an automated solutions which can be integrated easily and provides business manage incoming invoices easily and prevent any kind of fraud.

TrustBill Deck: https://vimeo.com/1096026701

Problem Statement:

Traditional invoice processing systems struggle to detect:

• Invoices from fake or unknown vendors
• Compromise of business email lead to tfaudulent bank account changes
• Duplicate invoices
• Unusually high or sudden payment requests go unnoticed
• Invoices missing clear, itemized details

Manual reviews are time-consuming, error-prone, and often ineffective against sophisticated scams that exploit these vulnerabilities.

What it does

TrustBill automates invoice fraud detection by:

1. Parsing incoming PDF or image invoices using AWS Bedrock Vision
2. Running them through a rule-based detection engine
3. Flagging invoices based on five proven fraud indicators:

Flagged invoices are categorized for human review through a feedback mechanism. If a flagged invoice is deemed legitimate, it’s added to the trusted vendor database, reducing future false positives.

How I built it

The application is completely serverless which uses multiple AWS serverless services like:

• AWS Lambda: for the compute, business logic and fraud detection engine
• API Gateway: to trigger lambda function, get data on the frontend side
• EventBridge: to trigger Verification process once all the details from the invoice is extracted
• S3 Bucket: to store the PDF files to review them again.
• AWS Bedrock: the vision capabillities are used to understand and extract data from variety of invoices.
• DynamoDB: to store the invoices and vendors Data

Accomplishments that I am proud of

In the end I am proud that I was able to complete the project in the given time.

What I learned

I learned how to architect a fully serverless solution using AWS services like Lambda, API Gateway, and DynamoDB while managing limitations like function timeouts. It was my first time using SAM which definitely made the process easier.

What's next for TrustBill

• Integrate AI-powered scam language analysis via Bedrock Claude for detecting urgent, manipulative phrasing in emails and invoices
• Build a dynamic, enterprise-configurable rules dashboard

How to test

In order to test this application you can send a PDF invoice on this email : [email protected]

Built With

• amazon-web-services
• aws-lambda
• lambda
• python
• serverless