Key takeaways
- Amazon Bedrock AgentCore enables organizations to deploy and operate secure AI agents at enterprise scale with seven core services.
- New offerings in AWS Marketplace help businesses find, buy, and deploy AI agents and tools from leading providers.
- AWS plans to invest an additional $100 million in the AWS Generative AI Innovation Center to boost agentic AI development and deployment.
- New Amazon Nova customization capabilities enable customers to build with higher accuracy and flexibility for their needs.
Amazon Web Services (AWS) has set the standard for security, reliability, and data privacy for cloud computing. Now, AWS is bringing these same principles to agentic AI with the announcement of new capabilities and tools that will help customers build AI agents on the solid foundation of AWS technology. Leading these innovations is Amazon Bedrock AgentCore, which will allow customers to deploy and operate highly capable AI agents securely at scale.
Swami Sivasubramanian, AWS VP for Agentic AI, led the keynote at the AWS Summit in New York, outlining the company's agentic AI strategy. He said AI agents—the autonomous software systems that leverage AI to reason, plan and adapt to complete—will dramatically accelerate innovation and improve productivity across every industry.
“It’s a tectonic change in a few dimensions,” Sivasubramanian said. “It upends the way software is built. It also introduces a host of new challenges to deploying and operating it, and potentially most impactfully, it changes how software interacts with the world—and how we interact with software.”
Deploy and operate AI agents at scale with Amazon Bedrock AgentCore
AgentCore is a complete set of services to deploy and operate highly effective AI agents using any framework and model.
As organizations race to adopt AI agents, they face a critical challenge: building systems that can act autonomously across digital boundaries, while maintaining the security, reliability, and governance standards required for enterprise deployment.
AgentCore helps developers bridge the critical gap between proof of concept and production for AI agents. It delivers a set of composable solutions that allows organizations to move agents from prototypes to applications that can scale to millions of end-users. Customers like Itaú Unibanco, Innovaccer, Boomi, Epsilon, and Box are already building with AgentCore.
Amazon Bedrock AgentCore services include:
AgentCore Runtime—Agents need a runtime that's both secure and dynamic enough to handle variable workloads. AgentCore Runtime supports interactive experiences with low latency and complex asynchronous workloads running up to eight hours, the longest in the industry. It is also the only framework agnostic offering that provides complete session isolation.
AgentCore Memory—Just as humans rely on both short-term and long-term memory, agents will rely on complex memory infrastructure to operate efficiently. AgentCore Memory makes it easy for developers to build context-aware agents by providing industry-leading long-term and short-term memory accuracy.
AgentCore Identity—Agents need to be able to securely access tools and resources to fulfill user requests using the right authentication, and AgentCore Identity provides seamless and secure agent authentication, integrating with existing identity providers such as Amazon Cognito, Microsoft Entra ID, and Okta.
AgentCore Gateway—AI agents need access to a wide range of tools to perform real world tasks, and AgentCore Gateway provides a secure way for agents to discover and use tools along with easy transformation of APIs, Lambda functions, and existing services into agent-compatible tools.
AgentCore Code Interpreter—AI agents need to write and execute code securely in sandbox environments to perform complex calculations, validate reasoning, process data, or generate visualizations. AgentCore Code Interpreter enables developers to customize environments with specific instance types and session properties to meet security requirements.
AgentCore Browser Tool—The model agnostic AgentCore Browser tool provides a fast, secure, cloud-based browser to enable AI agents to interact with websites at scale for tasks like form completion or navigating a website.
AgentCore Observability—Being able to track and trace every action of an agent is important for performance in production environments, and AgentCore Observability powered by Amazon CloudWatch gives developers real-time visibility through built-in dashboards and telemetry for key metrics—while integrating with existing observability systems.
New AWS Marketplace category will simplify enterprise AI agent adoption
At the AWS Summit, Sivasubramanian unveiled AI Agents and Tools in AWS Marketplace, which allows customers to discover, buy, deploy, and manage AI agents and tools from leading providers.
With AI Agents and Tools in AWS Marketplace, customers have streamlined search access to a one-stop shop for AI agent solutions and tools, allowing them to fast-track AI initiatives. Customers using AI Agents and Tools will build faster with these ready-to-integrate solutions, developing their AI strategies with professional services that specialize in building, maintaining, and scaling agents.
Doubling down on the AWS Generative AI Innovation Center
To accelerate customers’ development of autonomous, agentic AI systems, AWS announced it is making a second $100 million investment in the AWS Generative AI Innovation Center. The funding builds on two years of the center empowering thousands of customers around the world to boost productivity and transform their customers’ experiences.
Warner Bros. Discovery Sports Europe developed an AI-powered solution to help bike racing commentators research facts quickly using Amazon Bedrock, Anthropic's Claude 3.5, and other AWS services, and BMW built an AI solution on AWS that transforms how they diagnose network issues for over 23 million connected vehicles. Companies like Syngenta and AstraZeneca have also seen transformative results with agentic AI.
The center’s global team of AI scientists, strategists, and engineers works directly with customers and partners like these to solve the most complex challenges in AI implementation.
More news from AWS Summit in New York
New Amazon Nova customization capabilities enable customers to build with higher accuracy and flexibility for their needs
A powerful suite of customization capabilities is now available for Amazon Nova models on Amazon Sagemaker AI, with support for SageMaker HyperPod. This will allow customers to further adapt Nova models for their own needs. Also, enhancements to the Nova Act SDK will now bring additional security and access controls, helping developers take their prototypes to production and build agents that can reliably take action in web browsers.
Amazon S3 Vectors preview launches
Amazon S3 Vectors is the first cloud object storage with native vector support for AI workloads. With S3 Vectors, customers can reduce the cost of storing and querying vectors by up to 90% compared to conventional methods, making it cost-effective to retain and use large vector datasets to enhance AI as well as semantic search results of S3 data. It integrates with Amazon Bedrock Knowledge Bases and OpenSearch Service to streamline and reduce the cost of RAG and vector search operations.
Kiro bridges AI prototyping and production code
Kiro is a new IDE (Integrated Development Environment) for developers that simplifies the developer experience when working with AI agents on software development projects. With Kiro, developers can vibe code or go beyond with features like specs and hooks that allow them to collaborate with agents on a detailed plan and automatically trigger agent workflows to help with tasks like running tests or generating documentation.
New MCP resources will make it easier to build in the agentic world
Model Context Protocol (MCP) is an emerging standard that helps agents connect to data sources, tools, memory banks, and more. A new local AWS API MCP server contains complete knowledge of the full AWS API surface, allowing developers to vibe code with AWS and making it easier for any software assistant to become an AWS expert. A second resource, the AWS Knowledge MCP server offers an always-up-to-date MCP with comprehensive knowledge of AWS docs. It’s fully managed, continuously updated, and accessible remotely from any MCP client of your choice.
TwelveLabs AI models are now available in Amazon Bedrock
TwelveLabs AI transforms video libraries into searchable, actionable assets by processing visual, audio, and text elements simultaneously. Customers can now discover and work with content near instantly through natural language prompts, all underpinned by the security, privacy, and performance of AWS. AWS is the first cloud provider to offer models from TwelveLabs.
AWS and Meta team up for startups
AWS and Meta have joined forces to help startups build AI applications on Llama, offering 30 North American startups up to $200,000 in AWS credits each and technical support from Meta and AWS experts. Open for applications until August 8, 2025, this six-month program will accelerate AI innovation by helping startups build cutting-edge AI applications with Llama models.
Strands 1.0 makes it easier for AI agents to work together
AWS announced updates to the open source SDK Strands Agents, a new developer tool that makes it dramatically easier to create AI systems where multiple AI agents work together to solve complex problems. The technology reduces what previously took months of complicated technical work into a process that takes just hours, allowing businesses to build coordinated teams of AI assistants that can handle customer service, data analysis, and other complex tasks.
AWS AI League uses gamified challenges to help developers acquire essential AI skills
AWS launched the AWS AI League where developers can compete to solve real-world challenges with generative AI, while learning skills essential for innovating within their organization. The program offers up to $2 million in AWS credits for developers to get hands-on experience in fine-tuning, model customization, and prompt engineering. Top performers win all-expense-paid trips to AWS re:Invent and cash prizes.
AWS prepares early-career professionals for AI-transformed tech careers
AWS is providing students at more than 6,600 AWS Academy institutions free access to AWS Skill Builder’s subscription-tier training content and AWS Certification exam vouchers to build essential AI skills. Non-Academy students and recent graduates can access research that identifies promising opportunities and the recommended free learning plans in Skill Builder to pursue them. AWS has a goal to engage with 2.7 million students and early-career professionals globally within the first year.
Learn more about these launches and resources:
- Swami Sivasubramanian, AWS VP for Agentic AI,: Enabling customers to deliver production-ready AI agents at scale
- Introducing Amazon Bedrock AgentCore: Securely deploy and operate AI agents at any scale (preview)
- AWS doubles investment in AWS Generative AI Innovation Center
- New Amazon Nova customization capabilities
- Amazon S3 Vectors: First cloud storage with native vector support at scale (preview)
- Transform your development workflow with Kiro
- TwelveLabs video understanding models are now available in Amazon Bedrock
- AWS Startups: Building with Llama
- Strands Agents 1.0: Production-Ready Multi-Agent Orchestration Made Simple
Trending news and stories
