AWS unveils new agentic AI tools, managed services, and infrastructure at New York Summit

At the AWS Summit in New York City, Swami Sivasubramanian, AWS VP of Agentic AI, outlined a series of updates aimed at embedding agentic AI into enterprise workflows, security, and infrastructure. The announcements spanned new managed services, developer tools, and hardware accelerations, positioning AWS as a central platform for building and governing AI agents at scale.

Amazon Bedrock AgentCore received several updates to streamline agent development and governance. New capabilities include connectors for organizational, web, and paid knowledge sources, tools to diagnose production issues, and scalable controls for agent behavior as capabilities expand. A managed knowledge base for enterprise RAG pipelines was also introduced, offering native data connectors, Smart Parsing for multi-format data preparation, and an Agentic Retriever for complex queries—all integrated with AgentCore Gateway to reduce infrastructure management overhead.

AWS introduced a fully managed web search tool for Amazon Bedrock AgentCore, enabling agents to ground responses in current, cited web knowledge without data egress from secured AWS environments. This tool is designed to simplify agent development by eliminating the need for manual web search integration and infrastructure management.

Security and governance were emphasized with the announcement of AWS Continuum, a service in gated preview that prioritizes and remediates code vulnerabilities based on business impact. AWS Security Agent, now part of AWS Continuum, adds threat modeling using the STRIDE framework, pull request code scanning with remediation, and IDE integrations via Kiro power, Claude Code plugin, and MCP. These tools aim to reduce context switching for developers while improving security posture.

Developer experience was a focus with the introduction of Kiro for iOS, a native app in gated preview that allows engineers to initiate, monitor, and approve Kiro sessions directly from their phones. This extends Kiro’s capabilities beyond desktop environments, enabling remote oversight of agentic workflows.

AWS also announced AWS Transform, a continuous modernization service in preview that scans code repositories against configurable baselines to generate prioritized findings and autonomously create pull requests for remediation. AWS DevOps Agent adds release management capabilities, including release readiness reviews and autonomous release testing to verify changes against natural language standards before production deployment.

Infrastructure updates included the launch of Amazon EC2 G7 instances, accelerated by NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs. AWS claims these instances deliver up to 4.6x AI inference performance and up to 2.1x graphics performance compared to G6 instances, marking AWS as the first major cloud provider to support this GPU in the cloud.

Additional announcements included Amazon S3 annotations, which allow up to 1 GB of rich, queryable context to be attached directly to objects for AI agents and autonomous workflows, and Strands Agents, an open-source toolkit with improved context management, isolated execution environments, and chaos testing capabilities for production-grade agent development.