Some details
The US based space rocket manufacturer currently manually process thousands of invoices per month ranging from 5 line items to upwards of 10k line items.
The current system involves teams of supply chain experts reviewing many different types of document to ensure a high degree of accuracy can be maintained. Also, ITAR (International Traffic in Arms Regulations) compliance required that some ERP systems are
separated and kept in secure networks.Subsequently, considerable data needed to be synced to multiple systems. As multiple ERP systems are in use, the data flow would need to integrate with many different applications.
Appliscale investigated many off the shelf solutions.
Most off the shelf offerings were not fit for purpose due to ITAR requirements. Appliscale and the client agreed to develop a bespoke solution using the AWS GOV cloud. A number of its machine learning and character recognition capabilities would be utilized. The application would be integrated with existing ERP systems to achieve full invoice processing automation.
We designed a serverless application that is highly scalable and can be quickly deployed within minutes of feature updates being completed. With AWS serverless, all infrastructure is represented as code and fully automated build and deployment process was developed.
Due to the serverless architecture, the client doesn’t need to pay for server maintenance, only for code execution. AWS Lambda, CloudFormation, EC2, Sagemaker were utilized.
The next phase of the project is currently under preparation, this will see the evolution of the machine learning system to more accurately predict common errors and allow for human training of the system
Appliscale managed all of the technical requirements gathering, architectural design, development and project management.
