Amazon Bedrock Automation Streamlines Insight Extraction from Unstructured, Multimodal Content

AWS is announcing the general availability of Amazon Bedrock Data Automation, a new GenAI-powered capability of Amazon Bedrock that streamlines the process of extracting insights from unstructured, multimodal content, including documents, images, audio, and videos. Amazon Bedrock Data Automation (BDA) helps reduce development time and effort, allowing teams to more easily build intelligent document processing, media analysis, and other multimodal data-centric automation solutions, according to AWS.

Danilo Poccia, chief evangelist (EMEA), AWS, set the context for Amazon Bedrock Data Automation, explaining that, “Many applications need to interact with content available through different modalities…However, getting insights from unstructured multimodal content is not easy to set up: you have to implement processing pipelines for the different data formats and go through multiple steps to get the information you need.”

“That usually means having multiple models in production for which you have to handle cost optimizations (through fine-tuning and prompt engineering), safeguards (for example, against hallucinations), integrations with the target applications (including data formats), and model updates,” Poccia continued.

Amazon BDA aims to remediate this challenge, automating the generation of valuable insights from unstructured, multimodal content. Offering high accuracy at a low cost, as well as features for explainability and built-in hallucination mitigation, Amazon BDA delivers trustworthy, accurate insights from unstructured, multimodal data sources, according to the company.

Amazon BDA’s capacity to automate the transformation of unstructured content also yields significant benefit for building accurate GenAI apps, accelerating development processes by eliminating various time-consuming tasks—including managing multiple models, building complex orchestration pipelines, data preparation, and more. BDA’s seamless, managed experience—paired with its ease-of-use and customization features—empowers development teams to deliver business value, faster, according to AWS. BDA is also integrated with Bedrock Knowledge Bases, simplifying the way developers produce valuable information from their unstructured, multimodal content for retrieval-augmented generation (RAG) use cases.

AWS also enables customers to easily configure BDA’s outputs to produce specific insights in consistent formats as required by their unique business systems and applications. This customizability is achieved without sacrificing accuracy, further making auditing easy with features such as visual grounding and confidence scores.

Some AWS customers already see tremendous value in Amazon BDA for their GenAI and unstructured data needs:

“We are incredibly impressed with BDA and its ability to simplify the use of generative AI for our media and asset management needs, all while maintaining flexibility and accuracy,” said Philip Wisniewski, VP, global alliances, Tenovos. “With BDA, we can enable semantic search at scale, to increase content reuse by upwards of 50% or more and decrease millions of dollars in marketing costs. We also see tremendous opportunities in using BDA to automate visual quality assurance, reducing the need for manual inspection. We see a clear path for BDA to become a core component of Tenovos’ solution offering, driving real value for our joint customers. We’re excited to partner with AWS to leverage and enhance BDA as a transformative solution for the industry.”

“Our early look at AWS Bedrock Data Automation was promising,” said Samir Kazi, head of transformation and transitions, global business services, Clariant. “During our testing, it was effective at handling our complex document extraction needs across diverse formats. Its ability to consistently extract accurate data and adapt to various document structures stood out. This feature could greatly simplify our document processing and boost our operational efficiency.”

Amazon BDA is now generally available, currently offering support for cross-region inference in the U.S. East (N. Virginia) and U.S. West (Oregon) AWS Regions, with more to come.

To learn more about Amazon BDA, please visit https://aws.amazon.com/.