Hong Kong - December 9, 2024 - Amazon Web Services (AWS) announced multiple innovations for Amazon Bedrock at the 2024 re:Invent global conference. Amazon Bedrock is a fully managed service designed to build and scale generative artificial intelligence (GenAI) applications using high-performance foundational models.
The new features enhance AWS's commitment to model choice and optimize large-scale inference execution, enabling customers to derive more value from their data. The updates also help clients mitigate factual inaccuracies caused by model hallucinations, coordinate multiple AI-driven agents for complex tasks, and create smaller, task-specific models that deliver near-large model performance at lower costs and latencies.
Dr. Swami Sivasubramanian, AWS Vice President of Data and AI, stated, 'Amazon Bedrock continues to grow rapidly, becoming the choice for more customers due to its extensive model selection and built-in responsible AI features.' He emphasized the service's aim to address developers' challenges in harnessing generative AI's potential.
Customers can access over 100 popular, emerging, and specialized models through the Amazon Bedrock Marketplace. This includes models from leading AI companies such as AI21 Labs, Anthropic, and Stability AI. The newly released Amazon Nova model is exclusively available through Amazon Bedrock.
Amazon Bedrock Data Automation allows clients to convert unstructured multimodal data into structured formats for generative AI and data analysis. This feature streamlines the extraction and transformation of data from various content types, enhancing operational efficiency.
Symbeo, a CorVel subsidiary, plans to utilize Amazon Bedrock Data Automation to automate data extraction from complex documents, improving claims processing speed. Meanwhile, Tenovos has reported a 50% increase in content reuse through semantic search, resulting in substantial marketing cost savings.
With the introduction of Amazon Bedrock Guardrails, clients can automatically verify the accuracy of inference responses, enhancing transparency and compliance with organizational standards. The service's new multi-agent collaboration capabilities allow for specialized agents to be assigned for specific project steps, improving accuracy and task efficiency.