Subscribe to our daily and weekly newsletters for the latest updates and exclusive content on leading AI advancements.
Amazon is expanding its already impressive AI services portfolio, with a series of new features announced at the AWS re:Invent 2023 conference in Las Vegas. During an extensive keynote session that lasted nearly 2.5 hours, Adam Selipsky, CEO of Amazon Web Services (AWS)—the world’s largest cloud provider in terms of revenue, customer numbers, and data storage—highlighted the continued progress in enhancing its platform over the past year.
For 2023, much of Selipsky’s presentation focused on AI. Notably, he introduced Amazon Q, a generative AI automation service for AWS cloud services, including development, analytics, and operations. Selipsky also revealed several new features now generally available in the Amazon Bedrock generative AI service.
He was joined by Nvidia CEO Jenson Huang to detail their companies’ collaboration. Shortly after Huang’s segment, Selipsky announced AWS’s own Tranium 3 silicon for AI training, emphasizing AWS’s support throughout the AI lifecycle with advanced infrastructure, models, and applications.
Amazon Q, envisioned as an enterprise AI assistant, is being deeply integrated across multiple AWS services, such as Amazon CodeWhisper and Amazon Connect. It can answer questions, generate content, and perform tasks, easing the process of architecting, troubleshooting, and optimizing workloads. Amazon Q will aid developers by addressing service-related queries and resolving issues. It will also enhance various applications and business tools, specifically aiding contact center agents through Amazon Connect.
One of Amazon Q’s initial capabilities is code transformation, which presently helps organizations migrate technologies. For instance, Amazon used it to upgrade 1000 Java applications to a modern version in just two days, a process that historically took months per application. Future updates will enable migrating .NET workloads from Windows to Linux, potentially offering substantial cost savings.
Selipsky also announced several incremental updates to the Amazon Bedrock service. These include new customization features that enable customers to fine-tune AI models using their data, introducing capabilities like Retrieval-Augmented Generation (RAG) and continued pre-training. Bedrock now includes agents that facilitate actions such as booking travel or deploying software by orchestrating various organizational systems.
Additionally, Amazon Bedrock Guardrails have been introduced to enhance model safety, allowing customers to configure models to avoid certain topics or responses per their AI policies. Selipsky emphasized that this approach fundamentally redefines how organizations can build with generative AI.