記事数の多い日で、エージェント型 AI と Amazon Bedrock AgentCore が最大のテーマでした。AgentCore Payments によるガードレール付き決済、AgentCore を大規模運用する AgentOps、AgentCore Gateway の MCP サポート拡張、Policy/Lambda インターセプターによるエージェント保護、Identity での AWS Secrets Manager シークレット参照などが公開されました。同日 What's New でも AgentCore Identity の独自シークレット持ち込みが発表されています。OpenAI モデルと Codex の Bedrock 一般提供に関する導入ブログも登場しました。基盤サービスでは Amazon EC2 M8i/M8i-flex（ニュージーランド）や M8azn（アイルランド）の提供拡大、SageMaker HyperPod の EFA 専用ネットワークインターフェイス対応、Amazon Location Service の公共交通・複合経路案内、AWS Direct Connect の VIF レートリミッターなど幅広い更新がありました。日本語ブログは AWS Summit Japan/Tokyo 2026 の Database セッション紹介や AI-DLC の取り組みを扱い、セキュリティでは Spring 2026 の SOC 1/2/3 レポート（188 サービス対象）が公開されました。
エージェントAI/AgentCore: Payments・AgentOps・Gateway の MCP 拡張・インターセプター・Identity 連携
生成AIモデル: OpenAI モデルと Codex の Amazon Bedrock 一般提供の導入解説
コンピュート/インフラ: EC2 M8i・M8azn の地域拡大、SageMaker HyperPod の EFA 専用 NIC 対応
ネットワーク/サービス: Amazon Location の公共交通ルーティング、Direct Connect VIF レートリミッター
コンプライアンス: Spring 2026 SOC 1/2/3 レポート（188 サービス対象）公開
日本語コミュニティ: AWS Summit Japan/Tokyo 2026 の Database セッションと AI-DLC の取り組み
AWS Direct Connect now supports Virtual Interface (VIF) Rate Limiters on dedicated connections, which help you prevent network congestion caused by unexpected traffic spikes on a VIF which can potentially consume all available bandwidth, impacting workloads on other VIFs on the same connection.
With VIF Rate Limiters, you can set a maximum bandwidth allocation for up to 10 VIFs on a dedicated connection, choosing from a wide range available capacity increments from 50 Mbps to 1.6 Tbps when using a link aggregation group. Rate limiting applies to traffic both ingressing and egressing the AWS network. If traffic on a rate-limited VIF exceeds the configured capacity, excess packets are dropped, preventing that VIF from consuming bandwidth needed by other VIFs on the same connection. A new traffic utilization metric presented as percentage of the VIF’s configured capacity and dropped packet counts are published to Amazon CloudWatch, where you can configure alarms based on your thresholds. The new metrics make it easy to understand how your VIFs are using their bandwidth allocation and adjust accordingly.
VIF Rate Limiters are available in all AWS Regions in the commercial and China partitions where AWS Direct Connect dedicated connections are supported. You can configure Rate Limiters through the AWS Direct Connect console, API, or SDK.
To learn more, see VIF Rate Limiters in the AWS Direct Connect User Guide.
Amazon Bedrock AgentCore Identity now allows customers the ability to reference existing AWS Secrets Manager secret ARNs directly in AgentCore Identity Credential Providers.
Previously, AgentCore Identity used a service-managed secret approach, where secrets were created and managed by the service on the customer's behalf. This approach prevented customers from applying resource tags on create, encrypting secrets with a customer-managed key (CMK), or applying other organization-specific governance controls at the time of secret creation — causing friction for teams with strict governance requirements.
Now, customers create and manage their secrets in AWS Secrets Manager using their own governance and compliance policies, including custom CMKs, tagging strategies, automatic rotation and resource policies, and then reference the existing secret ARN when configuring a Credential Provider in AgentCore Identity. This gives customers full ownership of how their secrets are created, classified, and governed, without changing how AgentCore Identity uses them at runtime.
Amazon Bedrock AgentCore Identity bring your own secret is now generally available in 14 AWS Regions: US East (N. Virginia), US East (Ohio), US West (Oregon), Canada (Central), Asia Pacific (Mumbai), Asia Pacific (Seoul), Asia Pacific (Singapore), Asia Pacific (Sydney), Asia Pacific (Tokyo), Europe (Frankfurt), Europe (Ireland), Europe (London), Europe (Paris), and Europe (Stockholm). To learn more, visit the Amazon Bedrock AgentCore Identity documentation.
Starting today, Amazon EC2 M8i and M8i-flex instances are now available in Asia Pacific (New Zealand) Region. These instances are powered by custom Intel Xeon 6 processors, available only on AWS, delivering the highest performance and fastest memory bandwidth among comparable Intel processors in the cloud. The M8i and M8i-flex instances offer up to 15% better price-performance, and 2.5x more memory bandwidth compared to previous generation Intel-based instances. They deliver up to 20% better performance than M7i and M7i-flex instances, with even higher gains for specific workloads. The M8i and M8i-flex instances are up to 30% faster for PostgreSQL databases, up to 60% faster for NGINX web applications, and up to 40% faster for AI deep learning recommendation models compared to M7i and M7i-flex instances.
M8i-flex are the easiest way to get price performance benefits for a majority of general-purpose workloads like web and application servers, microservices, small and medium data stores, virtual desktops, and enterprise applications. They offer the most common sizes, from large to 16xlarge, and are a great first choice for applications that don't fully utilize all compute resources.
M8i instances are a great choice for all general purpose workloads, especially for workloads that need the largest instance sizes or continuous high CPU usage. The SAP-certified M8i instances offer 13 sizes including 2 bare metal sizes and the new 96xlarge size for the largest applications.
To get started, sign in to the AWS Management Console. For more information about the new instances, visit the M8i and M8i-flex instance page or visit the AWS News blog.
Starting today, Amazon EC2 M8azn instances are now available in Europe (Ireland) Region. These general purpose high-frequency high-network instances are powered by fifth generation AMD EPYC (formerly code named Turin) processors and offer the highest maximum CPU frequency, 5GHz in the cloud. M8azn instances offer up to 2x compute performance compared to previous generation M5zn instances, and up to 24% higher performance than M8a instances.
M8azn instances deliver up to 4.3x higher memory bandwidth and 10x larger L3 cache compared to M5zn instances allowing latency-sensitive and compute-intensive workloads to achieve results faster. These instances also offer up to 2x networking throughput and up to 3x EBS throughput versus M5zn instances. Built on the AWS Nitro System using sixth generation Nitro Cards, these instances are ideal for applications such as real-time financial analytics, high-performance computing, high-frequency trading (HFT), CI/CD, intensive gaming, and simulation modeling for the automotive, aerospace, energy, and telecommunication industries. M8azn instances are available in 9 sizes ranging from 2 to 96 vCPUs with up to 384 GiB of memory, including two bare metal variants.
To get started, sign in to the AWS Management Console. For more information visit the Amazon EC2 M8azn instance page.
Amazon SageMaker HyperPod now supports EFA-only network interfaces for cluster instance groups, enabling you to configure dedicated Elastic Fabric Adapter (EFA) devices without the traditional Elastic Network Adapter (ENA) for IP networking. SageMaker HyperPod is a purpose-built infrastructure for AI/ML model development that provides a resilient, high-performance environment with built-in fault tolerance and automated cluster recovery. Now with EFA-only, you can scale AI/ML clusters further without risking IP address exhaustion in your VPC.
When running large-scale distributed training workloads, inter-node communication bandwidth is critical to training performance. SageMaker HyperPod cluster instances support multiple EFA-capable network interfaces, but configuring them with the standard efa interface type attaches both an EFA device and an ENA device (for IP networking) to each interface — even when IP networking is only needed on a subset of interfaces within a node. The efa interface type inescapably consumes IP addresses in your subnet for each ENA device attached, which can lead to IP address exhaustion and limit the number of nodes you can deploy within a single subnet. With this launch, you can now set efa-only when configuring network interfaces for your HyperPod cluster instance groups. This option allocates the network interface exclusively for EFA traffic without attaching an ENA device, allowing you to maximize the number of EFA interfaces dedicated to low-latency, high-throughput inter-node communication. Because EFA-only interfaces do not require IP addresses, you can scale to larger clusters within the same subnets without encountering IP exhaustion. This configuration is particularly beneficial for large-scale distributed training jobs where inter-node communication bandwidth is critical and dedicated IP networking on every interface is not required.
To enable EFA-only, specify efa-only in the ClusterNetworkInterface configuration when creating or updating your HyperPod cluster via the CreateCluster/UpdateCluster API. EFA-only is available in all AWS Regions where Amazon SageMaker HyperPod is supported. To learn more, see ClusterNetworkInterface in the Amazon SageMaker API Reference.
Amazon SageMaker HyperPod now provides troubleshooting skills that bring expert-level AI/ML cluster diagnostics directly into AI coding assistants such as Claude Code, Cursor, and Kiro. SageMaker HyperPod is a purpose-built infrastructure for developing, training, and deploying foundation models at scale. It provides a resilient and performant environment with built-in fault tolerance, and automated cluster recovery, reducing the undifferentiated heavy lifting of managing large-scale AI/ML infrastructure. HyperPod skills enable you to diagnose and resolve cluster issues through natural language, reducing the time and expertise required to troubleshoot distributed training and inference infrastructure.
Debugging GPU hardware faults, diagnosing NCCL communication failures, and identifying performance bottlenecks across large distributed clusters remains complex and time-consuming. Operators often need to manually SSM into nodes, parse logs across dozens of instances, and cross-reference documentation. The new HyperPod troubleshooting skills help with faster time to resolution with capabilities spanning cluster health validation, hardware and communication diagnostics, software version drifts, and automated diagnostic reporting. Each skill encodes AWS best practices into structured diagnostic workflows that systematically guides AI agents to collect evidence from your cluster nodes via AWS Systems Manager, analyze patterns, and provide actionable recommendations. The skills work with your existing HyperPod infrastructure — no modifications are required.
The HyperPod troubleshooting skills are open source and available today for both Slurm and Amazon EKS orchestrated HyperPod clusters via the SageMaker AI skills plugin. To get started, visit the AWSLabs github repository to install the sagemaker-ai plugin in your preferred coding assistant.
Today, AWS Parallel Computing Service (AWS PCS) launches PCS-ready DLAMI, an AWS-maintained Amazon Machine Image built on the Deep Learning Base GPU AMI (Ubuntu 24.04). It provides a production-quality foundation for AI/ML training and high performance computing (HPC), with core infrastructure components pre-installed and tested for compatibility.
AWS PCS is a managed service that makes it easier for you to run and scale your HPC workloads and build scientific and engineering models on AWS using Slurm. You can use AWS PCS to build complete, elastic environments that integrate compute, storage, networking, and visualization tools. AWS PCS simplifies cluster operations with managed updates and built-in observability features, helping to remove the burden of maintenance. You can work in a familiar environment, focusing on your research and innovation instead of worrying about infrastructure.
The AMI inherits the operating system, NVIDIA GPU drivers, CUDA toolkit, EFA drivers, and Lustre client from the source Deep Learning Base GPU AMI, and adds PCS Agent, Slurm for PCS, and EFS utilities. Multiple supported Slurm versions are included, and the correct version activates automatically based on your cluster configuration. You can add frameworks, libraries, and application software on top to complete your environment. AWS releases updated AMIs regularly when the source DLAMI or PCS components are updated, providing ongoing security patches and driver updates.
AWS PCS-ready DLAMI is available for x86_64 and arm64 architectures at no additional cost in all AWS Regions where AWS PCS is available. To get started, specify a PCS-ready AMI when configuring your compute node groups. For more information, see Using PCS-ready DLAMI in the AWS PCS User Guide. For a reference cluster architecture that builds on PCS-ready DLAMI, see the awsome-distributed-ai repository on GitHub.
Amazon SageMaker Unified Studio now supports custom IAM permissions boundaries, so organizations that enforce Service Control Policies (SCPs) requiring permissions boundaries on all IAM roles can adopt SageMaker Unified Studio without modifying their security posture.
When a user creates a project, SageMaker Unified Studio provisions three IAM roles: a project user role, an Amazon Bedrock service role, and a Bedrock Lambda execution role. With this launch, administrators can specify a permissions boundary in the Tooling blueprint configuration, and all three roles are created with that permissions boundary attached. This satisfies SCP requirements at creation time, and project provisioning succeeds without administrator intervention. The permissions boundary also limits what the provisioned roles can do, so administrators retain control over project-level permissions even as new projects are created. Because the permissions boundary is set at the blueprint level, it applies to every new project automatically.
This feature is available in all AWS Regions where Amazon SageMaker Unified Studio is available. To learn more, visit the Manage Tooling blueprint parameters documentation.
Amazon Quick now enables enterprise customers to connect their privately hosted Model Context Protocol (MCP) servers to Quick through Amazon Virtual Private Cloud (VPC). Amazon Quick is an AI assistant that turns questions into answers, answers into actions, and actions into outcomes for you and your entire team. Previously, Quick's MCP support was limited to third-party hosted servers accessible over the public internet. With VPC support, organizations that host MCP servers on private networks for proprietary applications, custom data sources, and internal tools can now securely extend those capabilities to AI workflows in Quick.
With VPC connectivity for MCP, you can connect Quick to MCP servers running on Amazon EC2, AWS Fargate, AWS Agentcore, or other compute within your private network without exposing them to the internet. During MCP connector creation, select your VPC connection and provide your MCP server URL. Once connected, your team interacts with private MCP servers through natural language in Quick, with all traffic routed securely through your VPC.
VPC support for MCP servers is available in all AWS Regions where Amazon Quick is available.
Learn more about Amazon Quick and try for free. To learn more about connecting private MCP servers, visit the MCP documentation and the VPC connectivity guide.
Amazon Connect Customer now provides schedule update notifications, making it easier for you to automatically notify agents when their schedules change. You can define rules to send email or text notifications (via EventBridge) to supervisors and agents when new schedules are published, when existing schedules are updated, as well as when an agent’s leave request has a change in status. For example, you can automatically email all agents when their schedules for next month become available. These automated notifications eliminate manual effort for schedulers to notify workforce when schedules are updated and for agents to continuously monitor their leave request status, improving both scheduler and agent productivity.
This feature is available in all AWS Regions where Amazon Connect Customer agent scheduling is available. To learn more about Amazon Connect Customer agent scheduling, click here.
Amazon Quick Research now enables customers to encrypt their data using customer-managed keys (CMK) through AWS Key Management Service (KMS).
This enhancement allows organizations with strict security and compliance requirements to manage their own encryption keys. With customer-managed keys, you gain enhanced security control and comprehensive audit capabilities through AWS CloudTrail integration. You can encrypt your data with your own KMS keys, trace all data access for security auditing, and revoke access to compromised keys within 15 minutes during security incidents. This feature supports multiple CMKs with one default key per AWS account per region, providing the flexibility to manage encryption across different datasets while maintaining granular control over your sensitive business intelligence data.
Customer-managed keys must be created in the same AWS account and region as your Quick resources, and only symmetric AWS KMS keys are supported.
This feature is generally available in all AWS Regions where Amazon Quick is available. To learn more, visit the Amazon Quick Research detail page.
Amazon Connect Customer now supports up to 5,000 agents per schedule, making it easier for you to schedule larger business units or multiple business units that share agents (multi-skilled agents) within a single schedule. Additional scale limit updates include up to 350 agents per staffing group and up to 300 staffing groups per forecast group (for a total of up to 5,000 agents per forecast group). This launch eliminates the need to split scheduling across multiple runs or maintain separate schedules for shared agent pools, thus reducing operational complexity and enabling more accurate schedule optimization across the entire workforce.
This feature is available in all AWS Regions where Amazon Connect Customer agent scheduling is available. To learn more about Amazon Connect Customer agent scheduling, click here.
Amazon Connect Customer now optimizes placement of ad-hoc activities in agent schedules, making it easier for you to schedule non-productive events while automatically minimizing impact to service level goals. For example, automatically find the optimal time within the next 2 weeks to schedule compliance training for 500 agents. When adding activities such as training or meetings, you can choose one of the following placement methods: (a) anytime within the shift; (b) within a specific time window, for example, 12:00pm–4:00pm; (c) relative to the shift, for example, 1 hour after shift start and 2 hours before shift end. Based on your selected placement method, system automatically identifies optimal time within your specified constraints to schedule the activity while minimizing impact to service levels. This launch eliminates the need for supervisors to manually look for optimal times to schedule these events, thus improving supervisor productivity and ensuring consistent service levels for end customers.
This feature is available in all AWS Regions where Amazon Connect Customer agent scheduling is available. To learn more about Amazon Connect Customer agent scheduling, click here.
AWS HealthOmics now allows customers to specify the Nextflow engine version at run time via the StartRun API, enabling customers to pin runs to a specific Nextflow version for controlled migration. With this launch, customers can select from supported Nextflow versions (22.04, 23.10, 24.10, 25.10, 26.04) through the new engine-settings parameter, giving explicit control at the point of execution. AWS HealthOmics is a HIPAA-eligible service that helps healthcare and life sciences customers accelerate scientific breakthroughs at scale with fully managed bioinformatics workflows.
Nextflow version pinning gives customers full control over when and how they adopt new engine versions. The run-time version override ensures that even when a workflow definition specifies a version via manifest.nextflowVersion in its config or profile, the StartRun API parameter takes precedence, enabling customers to test the same workflow across multiple engine versions without modifying workflow source code. Production workflows can remain on a validated engine version while development teams test newer versions in parallel, reducing the risk of unexpected behavior changes. This is valuable for regulated environments where pipeline validation is required before upgrading to a new engine version.
Nextflow version pinning at run time is now available for Nextflow workflow runs in all AWS HealthOmics regions: US East (N. Virginia), US West (Oregon), Europe (Frankfurt, Ireland, London), Israel (Tel Aviv), and Asia Pacific (Singapore, Seoul). To learn more, visit the Nextflow engine settings documentation.
AWS HealthOmics now supports Nextflow version 26.04, enabling customers to take advantage of new Nextflow features and enhancements: record types, the strict syntax parser, workflow output summaries, and agent logging mode. AWS HealthOmics is a HIPAA-eligible service that helps healthcare and life sciences customers accelerate scientific breakthroughs at scale with fully managed bioinformatics workflows.
The strict syntax parser, now enabled by default in Nextflow v26.04, helps customers save compute time and costs by enforcing strict linting, consistent block structures, and unambiguous scoping, catching issues during pipeline initialization rather than hours into workflows. Record types allow workflow developers to write workflows with meaningful data names rather than keeping track of order of tuple elements, making workflows more readable, and less error-prone. Workflow output summary in JSON format simplifies integration with downstream tooling. Agent logging mode provides structured, minimal output optimized for AI-assisted workflow debugging and development.
Nextflow v26.04 is now available in all AWS HealthOmics regions: US East (N. Virginia), US West (Oregon), Europe (Frankfurt, Ireland, London), Israel (Tel Aviv), and Asia Pacific (Singapore, Seoul). To learn more, visit the AWS HealthOmics Nextflow workflow definition specifics documentation.
Amazon Relational Database Service (Amazon RDS) for SQL Server launches Bring Your Own Media (BYOM) for Microsoft SQL Server. With BYOM, customers who migrate SQL Server applications from on-premises environments can adopt a managed database service on AWS and reuse their existing Microsoft SQL Server licenses, including Software Assurance, through Microsoft's License Mobility program.
Amazon RDS provides a managed SQL Server database service that lowers operating costs with features such as high availability, automated backups and monitoring. BYOM helps customers who currently run Microsoft SQL Server on-premises, on other clouds, or as self-managed SQL Server on Amazon EC2, and want to adopt Amazon RDS and reuse their existing Microsoft SQL Server licenses. They no longer have to incur the cost of additional Microsoft SQL Server licenses, or wait for existing license agreements to expire to adopt RDS. Amazon RDS for SQL Server BYOM is integrated with AWS License Manager so customers can track their Microsoft SQL Server license usage across their AWS environment for licensing compliance.
To learn more about how to set up RDS SQL Server database instances with BYOM, visit the Amazon RDS SQL Server User Guide. For BYOM pricing and regional availability, visit the Amazon RDS for SQL Server pricing page.
AWS today announced that AWS Cost and Usage Report 2.0 (CUR 2.0) provides new integration options with AWS Athena and AWS Redshift. This capability allows customers to analyze the data from their AWS CUR 2.0 in Amazon Simple Storage Service (Amazon S3) using standard SQL without building custom data warehouse solutions, bringing feature parity with CUR 1.0 integration options.
With this launch, when customers select Athena or Redshift integration, CUR 2.0 exports are automatically delivered in the optimal format (Parquet, GZIP) for the chosen query engine. Each export includes the supporting metadata and automation resources needed to get started quickly, such as infrastructure templates, table definitions, and data loading instructions, so customers can begin querying their cost data without manual configuration. As CUR 2.0 data refreshes periodically, updates are automatically reflected in the Athena or Redshift tables with no additional ETL required.
This feature is available in all commercial AWS Regions, except the AWS GovCloud (US) Regions and the China Regions.
To learn more about this feature, see AWS Data Exports and AWS Billing and Cost Management in the AWS Cost Management User Guide.
Amazon Location Service today announced support for public transit and intermodal routing in the Routes API. Developers can now use the CalculateRoutes operation with two new travel modes, Transit and Intermodal, to plan journeys that combine public transportation with walking, driving, taxi, and rental segments.
With public transit routing, applications can calculate point-to-point routes using buses, subways, trains, ferries, and other transit types, including walking directions to and from stops, departure and arrival times, and transit line details. Intermodal routing extends this by combining multiple transport types in a single route, supporting common patterns such as park-and-ride (vehicle plus transit), taxi-and-ride (taxi plus transit), and last-mile completion using a taxi or rental. These capabilities help builders deliver applications across mobility, logistics, employee commute, and urban planning use cases that depend on accurate multi-modal route calculation.
Public transit and intermodal routing are available in the following AWS Regions: US East (Ohio), US East (N. Virginia), US West (Oregon), Asia Pacific (Mumbai), Asia Pacific (Sydney), Asia Pacific (Tokyo), Canada (Central), Europe (Frankfurt), Europe (Ireland), Europe (London), Europe (Stockholm), Europe (Spain), and South America (São Paulo).
To get started, visit the Amazon Location Service Routes Developer Guide, the Transit Routing and Intermodal Routing documentation pages, or the CalculateRoutes API Reference.
OpenAI frontier models GPT-5.5 and GPT-5.4, and Codex, the OpenAI coding agent, are now generally available on Amazon Bedrock. Deploy frontier models on Bedrock's high performance inference engine with built-in security, governance, and pay-per-token pricing.
Amazon が開発した RuleForge は、エージェンティック AI を活用して脆弱性検出ルールを自動生成するシステムです。ルール生成エージェントとジャッジモデルを分離するアーキテクチャにより、誤検知を 67% 削減しつつ、従来の手動プロセスと比較して 336% 速くルールを生成・検証できるようになりました。CVE 開示から防御までのギャップを埋め、AWS のお客様のワークロード保護を強化する仕組みを解説します。
大学生向けのプログラミング学習コミュニティ「POSSE」では、実践的な開発スキルやチームワークの向上を目指して […]
本記事は 2026 年 05 月 20 日に公開された “Introducing ExtendDB […]
みなさん、こんにちは。ソリューションアーキテクトの池田、ポール、佐山です。 2026 年 5 月 18 日〜2 […]
日本最大の「AWS を学ぶイベント」、AWS Summit Tokyo が 2026年6月25日(木)、26日 […]
日本最大の「AWS を学ぶイベント」、AWS Summit Japan が 2026年6月25日(木)、26日 […]
Amazon Web Services (AWS) is pleased to announce that the Spring 2026 System and Organization Controls (SOC) 1, 2, and 3 reports are now available. The reports cover 188 services over the 12-month period from April 1, 2025–March 31, 2026, giving customers a full year of assurance. These reports demonstrate our continuous commitment to adhering […]
In this post, we show how to build a comprehensive scalable user search layer on top of Amazon Cognito using AWS Lambda, Amazon DynamoDB, and Amazon OpenSearch Service.
This post details how NYCBS partnered with Amazon Web Services (AWS) and AWS partner Pronetx (now part of Caylent) to migrate to Amazon Connect Customer, the AWS cloud contact center service. The migration delivered a 54 percent improvement in patient enrollment and transformed the way NYCBS connects with the patients who need them most.
In this post, we show you how Doczy.ai™ uses generative AI on AWS to automate contract intelligence at scale, transforming unstructured documents into structured, actionable insights, so organizations can automate critical business processes and unlock the full value of their data.
In this post, we walk through a practical implementation using KDB-X MCP server integration with Amazon Quick, demonstrating how traders and analysts can ask questions using conversational language and receive actionable insights from datasets. You can apply this same integration pattern across various domains, from financial market analysis to IoT sensor monitoring to DevOps performance dashboards, where you need to simplify access to time series insights.
If you’re iterating on deploying large language models (LLMs) on AWS GPU instances, you’ve probably noticed the larger the model to be loaded into GPU High Bandwidth Memory (HBM), the longer the painful wait until the GPUs are ready for inference. As models grow to hundreds of billions of parameters and GPU environments grow ever […]
When you build agentic AI solutions, you face unique operational challenges. Agents make unpredictable decisions, costs spiral unexpectedly, and debugging non-deterministic failures seems impossible. Agentic AI applications don't just execute predetermined workflows. They reason, adapt, and make autonomous decisions, and DevOps practices need to be adapted. That's where AgentOps comes in, the operational discipline for deploying, managing, and continuously improving AI agents in production.
In this post, we address several key risks that surface when designing an agentic payment system, and how to address them with the capabilities of AgentCore payments.
In this post, we use a lakehouse data agent to demonstrate how you can use Policy for deterministic access control and Lambda interceptors for dynamic validation. We then show how to combine Lambda interceptors and Policy to implement a geography-based access control which requires both dynamic validation and deterministic access control.
While deploying Model Context Protocol (MCP) servers in production, enterprises need fine-grained access control across servers, observability into which teams use which tools, security guarantees against data exfiltration, and centralized credential management, all at scale. Amazon Bedrock AgentCore Gateway sits between MCP servers and the clients that consume them, centralizing credential management, observability, and secure […]
GPT-5.5, GPT-5.4, and Codex are now generally available on Amazon Bedrock. Deploy them in production applications and agents today, on Bedrock’s high performance inference engine.
In this post, we walk through how to use Amazon Quick Research to integrate biomedical data sources for rare cancer research. The walkthrough uses pediatric sarcoma as the research domain and draws on publicly available datasets from PubMed and other open biomedical repositories. It covers the end-to-end workflow: defining a research objective, configuring data sources, reviewing the AI-generated research plan, running the investigation, and iterating on results using the revision and versioning system.
Today, we’re excited to announce the ability to reference a secret in AWS Secrets Manager for AgentCore Identity, so you can reference your own preconfigured secret from Secrets Manager and retain full control over how it is managed. With this ability, you can extend your organization’s existing secrets governance processes to AgentCore. You can provide an existing, preconfigured AWS Secrets Manager secret to use with your credential provider resources. You retain full control over its encryption configuration, rotation, replication, tags, and resource policies, just as you would manage other secrets in Secrets Manager. You can also choose a secret from another AWS account within the same AWS Region, though cross-Region secret sharing isn’t supported. This also supports secrets brought in through AWS Secrets Manager external connectors, enabling integration with third-party secret managers.
This post demonstrates how to implement Open Authorization (OAuth) Code flow as an inbound authorization mechanism for MCP servers hosted on Amazon Bedrock AgentCore Gateway. By the end of this guide, you will have a production-ready setup where each AI assistant request is authenticated with a valid user identity token issued from your organization’s identity provider.
Multi-Region Event-Driven Failover Architecture with Amazon EventBridge and Route 53 Event-driven architectures enable applications to respond to events in real-time, providing scalability and loose coupling between components. However, ensuring high availability across multiple AWS regions requires careful design of failover mechanisms. This post demonstrates how to build a resilient multi-region event-driven architecture using Amazon EventBridge, […]