Amazon announced Amazon Bedrock AgentCore, a platform for building self-improving document processing systems, on its blog. It is the company's latest advancement in machine learning software for business automation.
The solution addresses challenges with traditional intelligent document processing (IDP) systems, which require extensive configuration and break when layouts vary across vendors. Amazon Bedrock AgentCore uses AI agents to classify, extract, validate, and learn from business documents with minimal human intervention.
It is built on Amazon Bedrock and targets organizations dealing with complex document workflows, such as supplier invoices and insurance claims. Availability begins with a public blog post, initially for developers and enterprise customers.
"Amazon Bedrock AgentCore is a platform to build, connect, and optimize agents at scale, with any framework or model," said the blog post. The solution enables agents to automatically handle new document formats and reuse learned knowledge for future tasks.
The announcement follows IDC research showing that document-related challenges cost organizations $19,732 annually per information worker and reduce productivity by 21.3 percent.
Amazon did not say how much the solution will cost, and it raises questions about how well it handles highly variable layouts. The blog post concludes by showing how the system processes new document types without manual template configuration.
Source: awsml
