Inspiration
The project started out as a direct need for a cheap, scalable and secure tool to handle invoice generation and validation. There are many third party solutions that offer similar solutions but none are as cheap to run or easy to use as this solution.
My inspiration for the project stemmed mainly from the needs I saw myself as well as around me and wanted to solve this challenging problem in the market.
What it does
The solution is able to generate invoices based on json or csv data and validate incoming invoices based on specific parameters to prevent fraud and faulty invoices. In it's first iteration it is a fully serverless solution that is deployed through infrastructure as code and a CI/CD pipeline. It enables user self-service sign-up and sign-in through cognito integration.
The provided UI is a simple tool that can be used to upload invoice data for generation as well as key metrics, historical data and key configuration for the validation.
How we built it
The solution is a complete package that is self-sustained and deployable quickly and easily. It's serverless architecture ensures low cost and scalability. It utilizes Bedrock Agentcore with the strands framework and an Amazon Nova PRO model to ensure consistent and intelligent processing.
Core Infrastructure
- Amazon Bedrock AgentCore: Runtime for autonomous AI agents
- Strands Framework: Python framework for agent development
- Amazon Bedrock: Amazon Nova PRO model for intelligent processing
- AWS Lambda: Serverless compute for invoice processing
- Amazon S3: Storage for invoice files and generated PDFs
- Amazon DynamoDB: NoSQL database for invoice records and caching
External Integrations
- Hardcoded Values: Tax rates and currency exchange rates (for demo)
- ReportLab: PDF generation library
User Interface & Security
- AWS Amplify: React-based admin dashboard
- AWS Cognito: Authentication and user management
- Amazon API Gateway: Secure API endpoints
- Amazon CloudWatch: Monitoring and alerting ## Challenges we ran into The biggest challenge was the automation of Bedrock Agentcore resource deployment as the documentation is limited and cloudformation does not provide support yet. ## Accomplishments that we're proud of I am very proud of the overall result already and am looking forward to working on this project further. ## What we learned I have learned a lot about the newest SDKs and frameworks that can be utilized to achieve great repeatable results and can't wait for more. ## What's next for GlobalInvoiceAI The solution provides a great solution for small to medium organizations that need cheap and scalable solution to their invoicing need and I am planning to continue to add features and optimize the solution to achieve the best possible results.
Built With
- amazon-web-services
- bedrock
- cloudformation
- github
- javascript
- lambda
- python
- react
- shell
Log in or sign up for Devpost to join the conversation.