Prime bulletins of the AWS Summit in New York, 2026

Immediately on the AWS Summit in New York Metropolis, Swami Sivasubramanian, AWS VP of Agentic AI, offered the day’s keynote.

Right here’s our roundup of the largest bulletins from the occasion:

New in brokers clients create

We’re introducing new capabilities on Amazon Bedrock AgentCore: connecting AI brokers to organizational, internet, and paid information, serving to groups discover and repair what’s going mistaken in manufacturing, and imposing controls that scale as brokers develop extra succesful.

Collectively, these capabilities aid you construct extra succesful brokers sooner, govern these brokers with controls that scale, and enhance them repeatedly. To be taught extra, learn our weblog submit overlaying all the brand new options.

• Introducing Amazon Bedrock Managed Information Base for sooner, extra correct enterprise AI functions – You possibly can construct enterprise RAG pipelines with the managed Information Base on Bedrock. It offers native knowledge connectors, Sensible Parsing for computerized multi-format knowledge preparation, and an Agentic Retriever for advanced multi-step queries, all built-in with AgentCore Gateway so builders can concentrate on enterprise outcomes moderately than infrastructure administration.
• Saying Internet Search on Amazon Bedrock AgentCore: Floor your AI brokers in present, correct internet information – You should use a totally managed internet search device that allows brokers to floor responses in present, cited internet information with zero knowledge egress from buyer’ secured AWS surroundings. You possibly can concentrate on constructing brokers as an alternative of manually including internet search to brokers on Bedrock AgentCore and managing its infrastructure.
• AWS WAF provides AI visitors monetization functionality to assist content material house owners cost AI bots for content material entry – You should use a brand new Bot Management functionality that allows content material suppliers and publishers worth, meter, and gather fee from AI bots and brokers accessing their content material and APIs. AWS WAF now enables you to set a worth for that entry, settle for fee via third-party suppliers, and grant scoped entry straight on the edge.
• Amazon Bedrock AgentCore harness in now usually obtainable – You are able to do constructing and working production-grade AI brokers in minutes, with out coding orchestration loops, by defining your agent’s mannequin, instruments, abilities, and directions in configuration, with Bedrock AgentCore harness.
• Coming quickly: AWS Context – This can be a new service that routinely maps the relationships throughout your current knowledge right into a information graph and offers agentic search so AI brokers within the group can entry ruled knowledge relationships, enterprise guidelines, and area information at runtime.

New in brokers for securing

• Introducing AWS Continuum: Safety at machine velocity – AWS Continuum for code vulnerabilities, obtainable in a gated preview, takes findings from throughout your surroundings, prioritizes by enterprise affect, proves that are exploitable, and drives a repair via your individual course of.
• AWS Safety Agent (now a part of AWS Continuum) provides menace modeling, Kiro energy and Claude Code plugin, and extra – You possibly can generate the brand new menace modeling (preview) to grasp the complete context of your utility and establish threats with really useful mitigations utilizing the STRIDE framework. You can too use pull request code scanning with remediation throughout main Git platforms, and IDE integrations through Kiro energy, Claude Code plugin, and MCP, letting builders run safety evaluations and repair points with out context switching.

New in brokers for constructing

• Introducing Kiro for iOS – Kiro introduces a local iOS app, obtainable in a gated preview, constructed for actual engineering work that offers builders a brand new floor to kick off, monitor, steer, and work together with their Kiro classes straight from their cellphone. Which means now you can begin classes, test again after they’re accomplished, assessment diffs, and approve modifications all whereas staying linked to your work with no laptop computer working.
• AWS DevOps Agent provides launch administration capabilities to evaluate code modifications earlier than manufacturing – You should use a brand new launch readiness assessment of code modifications and autonomous launch testing. These new options confirm each change towards the pure language requirements you give to the DevOps Agent and run change-specific exams in production-like environments.
• Proactively scale back tech debt autonomously with AWS Rework – steady modernization – You should use steady evaluation (preview) to routinely scan your code repositories towards configurable baselines and generates findings in hours, not weeks. When you’ve recognized and prioritized findings, you’ll be able to configure autonomous remediations that generate pull requests for affected repositories routinely.

New in brokers for works

With the launch of Amazon Fast’s new autonomous brokers, you’ll be able to create brokers that work within the background with particular experience, tone, and entry to instruments. You possibly can create a finance agent to course of orders as they arrive in, or a gross sales agent monitoring interactions throughout your CRM, emails, and Slacks to proactively draft follow-ups, flag dangers, or suggest subsequent steps.

We’re additionally releasing a brand new exercise feed that’s tailor-made to how you’re employed. It consolidates e-mail, messaging, calendar, and duties right into a single prioritized view, learns which messages you all the time reply quick, which threads you skip, and what subjects drive your week.

To be taught extra, look the demo of Amazon Fast – AI Assistant.

Along with the keynote bulletins, we now have different vital launches this week:

• Amazon S3 annotations: connect wealthy, queryable context on to your objects – Amazon S3 now enables you to connect as much as 1 GB of wealthy, mutable, and queryable context on to your objects utilizing annotations, purpose-built for AI brokers and autonomous workflows that want to find, perceive, and act on knowledge at scale with out sustaining separate metadata programs.
• Amazon ECS declares sooner service auto scaling – Amazon ECS service auto scaling now detects and responds to load modifications sooner with help for prime decision (20-second) metrics and metric publishing optimizations. In AWS benchmarking exams, time to set off scale-out improved from 363 seconds to 86 seconds (76% sooner, 4.2x), and complete time to scale and provision new duties improved from 386 seconds to 109 seconds (72% sooner, 3.5x).
• Amazon EC2 G7 situations accelerated by NVIDIA RTX PRO 4500 Blackwell Server Version GPUs – AWS is the primary main cloud supplier to help NVIDIA RTX PRO 4500 Blackwell Server Version GPUs. G7 situations are accelerated by these GPUs with customized sixth-generation Intel Xeon Scalable processors, delivering as much as 4.6x AI inference efficiency and as much as 2.1x graphics efficiency in comparison with G6 situations.
• Strands Brokers introduces new capabilities – Strands is an open supply toolkit for constructing manufacturing brokers. You should use now higher context administration within the Harness SDK, a brand new remoted execution surroundings with Strands Shell, and chaos testing and purple teaming in Strands Evals.

Up to date on June 18, 2026 — Added new vital launches on June 18.