Amazon Bedrock, part of AWS's suite of services for AI, has launched several new features aimed at enhancing the security and operational efficiency of AI applications. These updates include the introduction of resource-based policies, managed entitlements, zero data retention rules, and support for AI models in AWS GovCloud to ensure compliance and data security.
The platform now supports resource-based policies, which enable centralized access control for SaaS providers managing multi-tenant AI applications. These policies allow distinct security configurations tailored to different tenant requirements, such as cross-account access and private cloud traffic.
Additionally, zero data retention policies allow organizations to enforce control over data handling after inference requests, ensuring that no prompts or outputs are retained post-processing. This is particularly crucial for industries requiring strict compliance with data protection regulations.
The managed entitlements feature simplifies model access management across multiple AWS accounts by eliminating the need for individual AWS Marketplace permissions. This change reduces the operational overhead for organizations utilizing third-party models, streamlining AI adoption with centralized subscription management.
Furthermore, Amazon Bedrock's support for NVIDIA Nemotron and OpenAI GPT OSS models in AWS GovCloud provides US government agencies enhanced AI capabilities with the necessary compliance and data security measures.
✨ This summary was generated by AI from the outlets' reporting listed below. It is not independently verified and may contain errors — check the original sources. How BrevFeed works →
One email each morning: the day's tech stories, clustered across outlets and summarized. No account needed.
One email a day. Unsubscribe in one click, any time.
Spend a few minutes, get the whole day. Every topic's top stories in one hands-free rundown — listen, watch, or read the transcript.
▶ Play today's briefNew every morning, and the back catalogue is archived by date.
Amazon Bedrock AgentCore payments now allow AI agents to autonomously purchase x402-enabled software, APIs, data, and content directly from AWS Marketplace. This integration enables agents to pay in USD via the developer's AWS bill, removing the need for stablecoin wallets or API keys, while maintaining existing spending limits and auditability through AWS CloudTrail and Cost Explorer.
Amazon Bedrock AgentCore harness now uses the open-source Strands harness, introducing new defaults, built-in tools, and lifecycle hooks. This update provides agents with a recommended model, task-oriented prompts, automatic prompt caching, proactive context management, and tools for planning, web retrieval, and filesystem access, enhancing agent capabilities and control.
Skyflow and Amazon Bedrock AgentCore Gateway now integrate to prevent sensitive data from being exposed during AI agent tool calls and within model contexts. Skyflow Detect acts as an interceptor within AgentCore Gateway, de-identifying data before it reaches tools and restoring it before agents process results, without requiring changes to agents or tools. This integration addresses data leakage risks when AI agents interact with external services and documents.
Airbnb uses a cost-per-task metric to evaluate and select foundation models for its internal coding agents running on Amazon Bedrock. This approach helps Airbnb manage costs and performance as new models are released and usage patterns change, complementing operational metrics from Amazon CloudWatch.
Amazon Bedrock AgentCore Identity now supports private certificate authorities (CAs), allowing agents to connect to self-hosted identity providers (IdPs) like Keycloak or PingFederate within private networks. This feature enables validation of IdP server certificates issued by an organization's private CA, with the CA certificate stored in AWS Secrets Manager or Amazon S3. The integration uses Amazon VPC Lattice for private connectivity to IdPs.
Amazon Bedrock AgentCore Gateway can now connect to private APIs and Model Context Protocol (MCP) servers within hub-and-spoke network architectures that require centralized inspection. This pattern allows traffic between the Gateway and private targets to pass through a central firewall, addressing common enterprise network requirements. The solution details the placement of Amazon VPC Lattice resource gateways and configuration decisions for network and application teams.
Amazon Bedrock AgentCore users can reduce token costs by up to 71% and latency by up to 80% in JSON workloads by implementing three strategies. These methods address the issue of large JSON payloads being repeatedly injected into the LLM context window during tool invocations, which increases costs and response times.
Amazon Bedrock AgentCore Runtime provides a secure serverless environment for AI agents, but securing the ingress path requires additional measures. This article outlines two architectural patterns for creating a dedicated access layer to authenticate requests and enforce authorization before traffic reaches the runtime.
Amazon Web Services detailed three OAuth configurations for authenticating AI agents running on Amazon Bedrock AgentCore when calling Snowflake Cortex AI tools. These patterns address how agents obtain and present OAuth tokens to the Snowflake-managed MCP server, providing options for per-user, corporate SSO, and shared service user authentication. The configurations aim to extend enterprise authentication controls into the agentic layer without requiring broad service account permissions or custom integrations.
AWS introduced SMART (Social Messaging Agent Runtime Toolkit), an open-source Python SDK and Terraform stack, to simplify building AI agents for WhatsApp using Amazon Bedrock AgentCore. SMART handles complex WhatsApp integration challenges, such as message processing, conversation memory, and outbound messaging rules, allowing developers to focus on agent logic.
Microsoft announced several new Azure capabilities at Ignite 2026, including Azure AI core for controlled AI infrastructure, Azure Cobalt 200 VMs, and Azure Catalyst for agent-led modernization. The updates also feature HorizonDB for PostgreSQL with AI and analytics integration, and Fabric apps for governed data applications.
LoanPro, a lending platform provider, developed a multi-agent AI assistant for loan servicing using Amazon Bedrock and AgentCore. This assistant provides agents with synthesized borrower context and next-step recommendations, reducing customer call times by up to 15%. The solution aims to improve service quality and agent efficiency for the lenders using LoanPro's platform.
ArisGlobal scaled its NavaX AI engine on Amazon Bedrock to automate adverse-event case processing for life sciences. The company optimized costs by reversing its caching strategy to prioritize the shared prompt prefix, reducing inference expenses despite growing prompt sizes and increased tenant usage.
Amazon Bedrock now includes AgentCore memory, a new capability that allows AI agents to retain conversational context and user-specific facts for more personalized and relevant responses. This feature integrates with existing knowledge bases, enabling agents to use past interactions and user attributes to refine searches and ground answers in permitted documents.
Duco, a financial services platform processing 20 billion transactions monthly, has transitioned from traditional machine learning models to Amazon Bedrock and Amazon Bedrock AgentCore. This shift allows Duco to automate financial data processes with agentic AI, reducing deployment time and onboarding costs while maintaining data isolation for its banking clients.
MoEngage developed Merlin AI, a set of AI agents, by integrating Amazon Bedrock AgentCore to provide secure, tenant-scoped access to customer data. This allows marketing and product teams to manage customer engagement campaigns through a single conversational interface, reducing the need to switch between multiple tools.
AWS published a guide detailing a multi-stage methodology to modernize mainframe Assembler applications to Java Spring Boot using Amazon Bedrock and AWS Transform. This process involves converting Assembler to an intermediate COBOL stage before transforming to Java, aiming to reduce modernization time and maintain business logic integrity.
Amazon Bedrock Managed Knowledge Base now allows customization of answers per tenant using retrieval instructions. This feature enables setting specific voice, format, retrieval guidance, and answer policies for different audiences sharing the same documents. Previously, users had to manage prompt customization externally or disable response generation for consistent answer styles.
Amazon Bedrock now supports prompt caching for open-weight models, allowing reuse of computed states for repeated prompt prefixes. This feature reduces input costs and response latency by reading common context from a cache instead of reprocessing it. Users can utilize implicit caching, explicit caching, or both, depending on the model and API.
NVIDIA NeMo Switchyard, an open-source library, helps AI agents select appropriate models for coding tasks on Amazon Bedrock. This integration allows developers to route requests based on cost per solved task, optimizing model usage for coding agents like Claude Code and OpenAI Codex CLI. The provided guide uses AWS CDK to deploy LiteLLM with the Switchyard routing plugin to evaluate routing strategies.
Amazon Bedrock AgentCore runtime V2 now uses snapshots to accelerate session startup times and improve consistency. This change initializes the container environment once, then restores a snapshot for each new session, reducing the overhead of repeated container image pulls and initialization.
Amazon Bedrock now provides built-in cost attribution that traces every inference call to the IAM principal that made it, integrating with AWS Cost and Usage Reports (CUR 2.0) and AWS Cost Explorer. This feature allows organizations to track per-user costs and team spend for AI tools, regardless of the specific tool used, by operating at the inference endpoint layer. The update helps FinOps and AI Platform leaders manage and budget for AI tool adoption by providing detailed cost visibility and control options.
Natera, a cell-free DNA testing company, developed SWARM (Systematic Workflow Analytics and Resolutions Manager), an agentic AI system for lab operations. SWARM uses Amazon Bedrock AgentCore to automate responses to lab instrument support tickets, reducing first response time by 130x and achieving 96.2% automated coverage. This system addresses the challenge of lost institutional knowledge and slow response times in maintaining precision lab equipment.
Amazon Bedrock users can manage inference costs by adjusting request-level parameters like cache duration, service tier, reasoning effort, and maxTokens. These settings influence the number of tokens processed and their billing rates, offering optimization opportunities beyond prompt tuning. The article details how these choices impact costs and provides methods for verification through response metadata and billing data.
AWS now allows organizations to track and cap Amazon Bedrock generative AI spend per IAM principal using native AWS features. This solution uses automatic IAM principal attribution, AWS Cost and Usage Reports (CUR) 2.0 or Cost Explorer for reporting, and AWS Budgets with budget actions for enforcement, eliminating the need for custom code in common scenarios.
This post details a framework for calculating true model cost on Amazon Bedrock, focusing on cost per task rather than cost per token. It also provides techniques for reducing costs through caching, routing, and output constraints, and a method for measuring agent ROI using human baselines and a five-dimension scorecard.
A guide details how to construct a self-service Large Language Model (LLM) gateway using Amazon Bedrock and LiteLLM on AWS Fargate. This setup abstracts Amazon Bedrock behind an OpenAI-compatible API, providing developers with simplified access to foundation models. The gateway enables centralized monitoring of latency, errors, and cost attribution, along with consistent access controls for platform, finance, and security teams.
Amazon Bedrock has released a migration guide within the Agent Toolkit for AWS to assist users in moving from customer-managed knowledge bases to the Amazon Bedrock Managed Knowledge Base. This guide helps automate the migration process, allowing coding agents to plan and create new knowledge bases, while Amazon Bedrock handles scaling, indexing, and retrieval.
AWS has demonstrated a reference architecture for building composable, interoperable multi-agent systems across organizational boundaries using the Agent-to-Agent (A2A) protocol. This allows agents built with different frameworks to discover, communicate, and exchange tasks securely without custom integration code or shared secrets, addressing a key challenge in multi-agent system scalability and collaboration.
Amazon Bedrock Managed Knowledge Base tools now offer fine-grained access control, allowing AI agents to restrict document searches based on user permissions. This feature addresses data security concerns in multi-tenant environments and internal assistants where agents access sensitive information on behalf of users.
Amazon Bedrock and Spring AI now allow Java developers to build autonomous AI agents that can plan, use tools, and maintain context in multi-turn conversations. This integration addresses scalability, governance, security, and operational challenges for bringing AI agents to production within established Java ecosystems. The combination enables Java teams to move beyond simple prompt-response interactions to complex multi-step tasks.
OpenAI Codex can now use Amazon Bedrock for model inference, allowing developers to integrate Codex with existing AWS identity, regional, and observability practices. This integration provides platform teams with options to manage model access and apply policy controls through various access paths like Native AWS Access or AgentCore Gateway.
Amazon Bedrock AgentCore launched a new Web Search API, allowing developers to directly integrate web search capabilities into their applications with a single API call. This API provides ranked search results from an Amazon-operated web index, offering an alternative to the existing AgentCore Gateway for scenarios requiring more direct control or custom orchestration.
AWS announced updates to Amazon Bedrock and AgentCore in October 2026, introducing new AI models like Claude Haiku 5.5 and OpenAI GPT-6.1 Terra, alongside features for agent reliability, process adherence, and cost management. These updates provide builders with more model choices for specific tasks, improved agent operational control, and better cost predictability for AI workloads.
AWS introduced an event-driven architecture to automate updates for the AWS Agent Registry, a component of Amazon Bedrock AgentCore. This architecture uses AWS CloudTrail, Amazon EventBridge, and AWS Lambda to synchronize changes to Model Context Protocol (MCP) server tools and schemas, preventing stale metadata in the registry. The solution addresses issues of outdated information and maintains trust in the catalog for organizations managing multiple MCP servers.
DXC Technology developed a governed multi-model AI substrate using Amazon Bedrock to manage AI agents for enterprise IT operations. This substrate provides access to various foundation models and enforces content filtering, PII redaction, and denied-topic policies at the infrastructure level. DXC OASIS, built on this substrate, has reduced resolution time by 95% across 56 use cases since its general availability in March 2026.
Amazon Bedrock AgentCore released its Web Search API, allowing AI agents to access current web results through a single API call within the AWS environment. This API provides agents with up-to-date information for tasks like monitoring security advisories or product releases, and integrates with LangChain and LangGraph for developers.
This post details methods for evaluating AI agents deployed on Amazon Bedrock AgentCore. It explains how to use AgentCore's built-in evaluators alongside DeepEval and AutoEval to measure agent performance against business objectives. The guidance focuses on defining success in business terms and mapping observable behaviors to appropriate evaluation metrics.
Trovata has updated its cash and liquidity management platform with autonomous AI agents built on Amazon Bedrock and AgentCore. These agents automate treasury tasks, moving beyond static insights to continuous workflows for corporate treasury teams. This development addresses challenges like fragmented banking data and manual reporting, aiming to improve real-time cash decisions.
Tata Consultancy Services (TCS) built an agentic test orchestrator with Amazon Bedrock AgentCore to improve testing throughput for its Banking, Financial Services, and Insurance (BFSI) clients. This system converts requirements into executable, governed tests, aiming to reduce repetitive tasks for quality engineers. The orchestrator addresses specific needs for project-specific context, API and UI testing, data isolation, and repeatable workflows.
AWS has released a centralized observability data lake for Amazon Bedrock model invocations, addressing the challenge of monitoring AI workloads across multiple accounts. This solution aggregates invocation logs, enabling users to track prompt details, token usage, model responses, and per-invocation costs. The data lake provides a single, queryable view of all invocations, which helps platform and machine learning engineers debug usage spikes and identify inefficient prompts.
Amazon Bedrock Managed Knowledge Base now supports single-shot Retrieval Augmented Generation (RAG) and retrieval-only requests via AgenticRetrieveStream. This update allows applications to perform a single search for direct questions or return raw passages, complementing the existing agentic retrieval for complex, multi-search queries. The new capabilities enable more efficient and flexible handling of diverse AI assistant queries within a single API.
AWS has released a Pluggable Agentic AI Framework, a seven-layer reference architecture for building and operating multi-agent AI applications on Amazon Bedrock AgentCore. This framework addresses challenges in scaling AI agents from prototypes to production, including security, reliability, and cost management, by providing a standardized, composable structure.
Amazon Bedrock AgentCore now includes multi-layer write validation to protect against memory poisoning in AI agent memory. This enhancement adds an embedding-based semantic-drift signal to existing controls, helping to identify content that deviates from session context or uses obfuscated phrasing. The update strengthens the integrity of AI agent memory by preventing malicious instructions from persisting across sessions.
CrossnoKaye implemented Amazon Bedrock AgentCore to integrate AI into industrial operations, addressing challenges posed by disparate systems and proprietary control logic. Its ATLAS platform connects plant data and AI reasoning to closed-loop control, managing facility models, control software, historians, and optimization applications within a single platform on AWS. This integration allows AURA, CrossnoKaye's agentic intelligence system, to access the full operational state of a plant and use Bedrock's code interpreter for numerical computations.
AWS released AgentCore Runtime V2, which optimizes memory usage and cold start times for agents, and introduced managed migration for Apache Hive tables to Apache Iceberg via Amazon S3 Tables. These updates aim to reduce operational overhead and costs for users by improving efficiency and simplifying data migration processes.
AWS launched Kiro bots, adding persistent agent memory and delegated workflow management to its Kiro development environment. This allows individual AI agents to retain context across sessions and manage parallel tasks, reducing the need for developers to manually track and integrate agent work. The update shifts the focus from delegating individual tasks to delegating entire processes, addressing issues like 'context rot' in multi-agent workflows.
Amazon's North America Sort Center (NASC) network developed a serverless Model Context Protocol (MCP) server using Amazon ECS with AWS Fargate. This server acts as a centralized gateway, allowing AI agents to access enterprise data lakes without duplicating integration logic or custom code for each dataset. The solution addresses challenges of fragmented data access and simplifies onboarding new data sources for AI applications.
This article details how to scale AI agent access governance using Okta Cross App Access (XAA) and Amazon Bedrock AgentCore. It describes implementing a single front door with Amazon Bedrock AgentCore Gateway and using Cedar policies for tool access, expanding on a previous solution.
Okta Cross App Access (XAA) integrates with Amazon Bedrock AgentCore to manage and enforce AI agent access to user applications. This solution allows administrators to define which agents can act on behalf of specific users within approved applications, with Okta handling policy evaluation and AgentCore Identity managing token issuance and storage. This integration addresses the challenge of controlling AI agent permissions in enterprise environments.
Greenlight Financial developed an autonomous Security Operations Center (SOC) solution named Luna using Amazon Bedrock AgentCore. Luna resolves over 90% of 3,000-4,000 monthly security and risk cases without human intervention. This implementation demonstrates how AI agents can automate security triage in regulated financial environments.
Taxbit, a digital-asset tax and accounting compliance platform, introduced Taxbit Assist, an AI assistant embedded in its product dashboard. This tool, built on Amazon Bedrock AgentCore, provides grounded and cited answers to user queries, improving access to product knowledge previously scattered across various sources. The assistant is designed to be role-aware, avoid tax advice, and adhere to regional data residency and compliance standards.
Amazon Bedrock AgentCore now allows users to run Claude Code as a remote agent, enabling offloading coding tasks to isolated sessions within an AWS account. This integration maintains Amazon Bedrock's data controls, including data protection, IAM-governed access, and zero operator access for customer data.
Amazon Bedrock AgentCore now offers Web Answer, a new feature that provides synthesized, source-cited answers to natural-language questions with a single tool call. This capability automates the process of searching, retrieving, synthesizing, and citing information from current web sources, reducing the need for developers to build and maintain complex pipelines for agents.
Amazon Bedrock AgentCore Evaluations now integrates decision models for agent scoring, allowing for more consistent and quantifiable evaluations. This feature enables users to define custom evaluators that connect to external decision model backends like Jev or Strands Decider 2B, providing calibrated probabilities for specific, repeatable questions.
U Mobile, a Malaysian 5G provider, developed AiDaH, an AI-driven agentic helpdesk, using Amazon Bedrock AgentCore to automate complex operational problem diagnosis and resolution. This system navigates multiple internal systems and knowledge bases, reducing trouble ticket volumes by up to 84.5% in key categories.
Amazon Bedrock AgentCore Gateway now supports on-behalf-of (OBO) token exchange with Microsoft Entra ID, allowing AI agents to call ServiceNow REST APIs while preserving the original user's identity. This integration ensures audit trails, enforces user-level access controls, and upholds the principle of least privilege for API interactions.
Gen Digital developed Norton Revamp, a professional reputation management platform, using Amazon Bedrock AgentCore Web Search. The platform's AI-curated News Engine provides real-time, cited web knowledge to inform content suggestions and news feeds for users. This integration allows Norton Revamp to offer current industry information without the overhead of custom crawlers or third-party APIs.
This post details how to construct a semantic layer for agentic AI using EDB Postgres AI, enabling a Strands Agents agent on Amazon Bedrock AgentCore to answer customer 360 questions. The method integrates business meaning directly into the operational Postgres database via an AIDB extension, utilizing standard Postgres features for data representation, join paths, and governance. This approach allows agents to query live data while maintaining security and context.
Amazon Bedrock now supports in-country inference in India for the Kimi K3 multi-modal foundation model, enabling local workloads with strict data residency requirements. This expansion allows customers to process data within India, addressing specific regulatory needs for applications like AI health assistants and document understanding.
Amazon Web Services introduced Crawl4AI, an open-source Python crawler, to prepare web content for Amazon Bedrock Knowledge Bases. This tool allows for custom content preparation, such as indexing specific page regions or adding metadata, before ingestion into Bedrock's RAG applications. It complements the existing Amazon Bedrock Web Crawler for scenarios requiring more control over data ingestion.
A study investigated how AI agents recommend tools by asking Claude and other engines about CMS platforms supporting MCP integration. Webflow appeared in 49% of 100 responses across four engines, with significant variation between engines. This highlights how AI agent recommendations influence developer tool discovery.
Amazon Bedrock AgentCore runtime now allows developers to preview web applications live in their browser as coding agents build them. This feature, called Session Endpoints, provides an authenticated URL to access apps running within an isolated microVM session, enabling real-time feedback and faster iteration cycles for agent-generated code.
The ChatGPT desktop app is now available on Amazon Bedrock in AWS GovCloud (US), offering access to OpenAI's models, including Codex, for organizations with specific security and data residency requirements. This integration allows model inference to be routed through AWS GovCloud (US) endpoints, enabling governance via AWS IAM and existing network controls.
Epsilon, a data and marketing-services company, adopted Amazon Nova Act and Amazon Bedrock AgentCore Browser to provide intent-driven UI testing as a managed service for its product teams. This integration allows Epsilon to scale UI testing across its applications without increasing maintenance overhead for individual teams, complementing their existing Playwright-based regression suite.
MoEngage integrated over 400 internal REST APIs into AI agent tools using Amazon Bedrock AgentCore's AgentCore Gateway. This approach allowed their AI agents to access existing functionalities securely without exposing private APIs to the public internet or requiring individual teams to build Model Context Protocol (MCP) servers. The integration saved an estimated six months of platform engineering effort and reduced API onboarding time from days to hours.
This article details the architecture and reference implementation for an enterprise Agentic AI Platform using Amazon Bedrock and Amazon Bedrock AgentCore. It addresses challenges in deploying AI agents in production, such as secure environments, ownership, and cost management, by providing a structured approach for platform and workstream teams.
This post details how to build enterprise agentic retrieval on a Managed Amazon Bedrock Knowledge Base using the AWS Cloud Development Kit (CDK). This approach addresses limitations of traditional RAG systems by incorporating a reasoning agent for multi-turn planning and tool-driven retrieval.
Google Cloud has launched IAM Helper, a new suite of tools designed to automate complex Identity and Access Management (IAM) workflows for autonomous agents. This offering integrates with existing Google Cloud IAM engines to allow agents to manage access dynamically, aiming to reduce administrative overhead and improve security for cloud resource provisioning and application deployment.
AWS released Continuum, a new service for end-to-end code vulnerability management that identifies, prioritizes, and suggests fixes for exploitable vulnerabilities. Additionally, Amazon Polly's Generative engine received an update, adding 14 new voices, including two for Brazilian Portuguese.
Bayer Crop Science developed the Agentic Decision Engine (ADE) using Amazon Bedrock AgentCore to provide a natural language interface for its Connected Decision Engine (CDE). This system allows users to query complex machine learning workflows without navigating multiple operational systems, resolving 80% of user questions without engineering intervention. The ADE addresses the bottleneck of understanding ML workflows by centralizing information access.
The Adaptive Agentic Intelligence Framework (AAIF) was introduced, utilizing Strands Agents to route AI requests. This framework employs small language models (SLMs) for local, routine tasks and larger cloud-hosted models via Amazon Bedrock for complex requests, aiming to reduce API costs and latency.
Amazon Quick and Amazon Bedrock AgentCore now provide conversational supply chain intelligence, allowing users to query multiple enterprise systems in plain language. This integration aims to reduce the time and complexity involved in gathering insights from disparate supply chain data sources, enabling faster decision-making.
Octave developed a secure, multi-tenant AI foundation on Amazon Bedrock, utilizing Cohere Embed v4, to process regulated data for quality management in industries like medical devices and pharmaceuticals. This foundation allows AI to analyze billions of documents while maintaining strict data isolation for over 14,000 customers, addressing the challenge of leveraging AI in highly regulated environments.
AWS Agent Registry is a new managed service that allows AI agents to discover tools and APIs at runtime, rather than having them hardcoded. This addresses scalability issues for agents interacting with large numbers of internal APIs in enterprise systems, reducing the need for constant code changes and redeployments.
Keyfactor EJBCA integrates with AWS services to secure AI agent identity using X.509 Public Key Infrastructure (PKI). This solution addresses the lack of built-in authentication in multi-agent AI frameworks like the Model Context Protocol (MCP) by providing cryptographic proof-of-possession and real-time revocation checking.
Eon implemented Amazon Bedrock AgentCore to create a single authenticated gateway for multi-cloud investigation tools across AWS, GCP, and Azure. This integration replaces individual tool credentials and local Model Context Protocol (MCP) processes, reducing setup time for new engineers from two days to two minutes.
AWS now supports building semantic layers for agentic AI applications using the Denodo Platform with Amazon Bedrock AgentCore. This integration allows AI agents to access and query distributed enterprise data, including on-premises PostgreSQL databases, Amazon Aurora, and Amazon Redshift, through Denodo's VQL dialect.
This post details how to construct a semantic layer for agentic AI on AWS using Strategy Mosaic, integrating with Amazon Bedrock AgentCore. It demonstrates modeling a customer 360 domain across Amazon Redshift and Amazon Aurora without ETL, utilizing Mosaic's dimensional model and metric expression syntax. This setup allows AI agents to query federated data sources through Mosaic's Model Context Protocol.
AtScale now functions as a semantic layer for agentic AI on AWS, integrating with Amazon Bedrock AgentCore. This integration allows AI agents to query dimensional semantic models authored in AtScale's Semantic Modeling Language, providing consistent data definitions for both AI and business intelligence tools.
Amazon Bedrock now supports structured outputs for all Anthropic Claude models on both bedrock-runtime and bedrock-mantle endpoints. This feature provides schema-compliant JSON responses, eliminating the need for custom parsing and retry logic in applications.
AWS launched a preview of its Well-Architected Agent, an AI service that analyzes cloud environments against Well-Architected best practices. The agent provides prioritized recommendations across cost, security, performance, and reliability, and can suggest remediation steps and review infrastructure-as-code before deployment.
Amazon developed a continuous quality evaluation framework for AI agents using Amazon Bedrock AgentCore. This framework addresses the issue of AI agents providing confident but incorrect answers, which can degrade trust and impact business decisions. The system aims to ensure agents maintain accuracy at scale across various sub-agents and workflows.
Amazon Bedrock AgentCore can be used to build agentic intelligent document processing (IDP) systems that classify, extract, validate, and learn from business documents. This approach allows agents to automatically handle new document formats and learn from corrections, reducing manual configuration and maintenance.
A new solution uses Amazon Bedrock AgentCore to automate the creation of client PowerPoint presentations from financial reports, reducing creation time to under 90 seconds. The system employs a supervisor agent to orchestrate document generation and summarization tools, integrating models like Anthropic Claude Sonnet 4.6 for content and Stability AI Stable Image Inpaint for visuals. This approach aims to streamline presentation workflows for financial services teams by grounding content in source documents.
Tamnoon developed Tami, an AI-powered cloud remediation orchestrator using Amazon Bedrock, to automate the investigation and remediation of cloud security findings. This solution addresses the challenge of overwhelming alert volumes and manual investigation in cloud security, aiming to reduce critical cloud exposures. Tami achieved over a 90% reduction in critical cloud exposure within 90 days without production issues.
Anchor Browser now integrates with Amazon Bedrock AgentCore payments, allowing AI agents to autonomously pay for web resources and APIs that return HTTP 402 Payment Required. This integration enables agents to continue workflows without manual intervention when encountering paywalled content or metered services.
AWS introduced managed entitlements for Amazon Bedrock, allowing organizations to manage private pricing offers for third-party AI models from a central account. This update simplifies multi-account access and automates the activation of negotiated pricing, reducing operational friction for AWS Organizations users.
Orca Security partnered with AWS to develop Discovery AI, a natural language query agent built on Amazon Bedrock. This integration allows security analysts to query cloud infrastructure data using plain text, reducing query costs by 90%, input tokens by 93%, and latency by 75% compared to previous methods.
ShipStation Global created an AI-powered IT help desk agent named Deckhand, built on Amazon Bedrock AgentCore, to automate responses to routine IT questions and manage Jira tickets within Slack. This system resolved 97.5% of incoming queries and achieved a 96% positive feedback rating from employees, freeing up IT staff time. The solution addresses the common IT challenge of repetitive questions and after-hours support by centralizing information access and automating responses.
AWS released updates for Amazon Bedrock, AgentCore, and Strands in September 2026, focusing on expanding model choice, improving agent operations, and connecting AI applications with business information. These updates include public preview of Bedrock Managed Agents for OpenAI models and enhanced AgentCore runtime efficiency.
Postman launched Agent Mode, an AI-native interface for its platform, leveraging Amazon Bedrock to serve 40 million developers. This integration addresses challenges in adapting a mature product for AI agents, focusing on architectural patterns like tool control and schema-based reads.
Amazon Bedrock AgentCore Payments now allows AI agents to pay for services like model inference on demand, with spending limits enforced by the infrastructure. This service simplifies integrating payment protocols for high-frequency, low-value transactions within agent loops, as demonstrated by Incarna using it to pay BlockRun for model inference.
Amazon Quick and Amazon Bedrock Knowledge Bases now offer real-time access control list (ACL) enforcement for Retrieval Augmented Generation (RAG) systems. This development addresses the challenge of ensuring AI-generated answers respect existing document permissions in enterprise knowledge sources.
Qlik developed Qlik Answers, an AI tool built on Amazon Bedrock, to enable enterprise employees to quickly query diverse data sources using natural language. This solution aims to provide verifiable answers with sources, addressing issues of data sovereignty and scalable AI model capacity for Qlik's 40,000 customers.
AWS published a guide on automating incident remediation after an AWS DevOps Agent investigation using AWS Lambda Durable Functions, Amazon EventBridge, and Amazon Bedrock. This workflow aims to reduce mean time to resolution by transforming investigation summaries into pre-validated fixes requiring single approval.
Cornerstone OnDemand developed Orion AI, a multi-agent AI system utilizing Amazon Bedrock and Strands Agents, to automate database operations. This system reduced database incident diagnosis time from 45 minutes to 10 minutes, a 78% improvement. Orion AI transforms reactive database management into proactive, self-orchestrating workflows for the company's Enterprise DataOps team.
A guide demonstrates how to build AI assistants that retain conversational context using OpenClaw, an open-source agentic system, and AgentCore runtime, a capability of Amazon Bedrock. This approach allows assistants to remember past interactions and apply that knowledge to future queries, moving beyond stateless, single-question responses.
AWS published a guide on implementing identity-aware AI data agents that respect existing governance policies in AWS Lake Formation. This approach allows AI agents to propagate user identities through the system, ensuring Lake Formation evaluates individual user grants without requiring policy rebuilding in application code.
AWS published guidance on how its services and frameworks can help customers align with the upcoming ISO/IEC 42005:2025 standard for AI system impact assessments. This initiative aims to assist organizations in integrating responsible AI governance practices into their operations, particularly for the "facilitator" role in AI adoption. The guidance highlights AWS tools that support the implementation of AI impact assessments as codified by the new ISO standard.
AWS published a tutorial detailing how to build a voice-enabled travel concierge for airline applications. The solution uses Amazon Bedrock AgentCore, Amazon Nova Sonic, and Amazon Bedrock Knowledge Bases to enable spoken requests for tasks like checking flights or changing seats.
Zhipu AI's GLM 5.3, a 753B-parameter mixture-of-experts model, is now available on Amazon Bedrock, offering enhanced capabilities for coding and long-horizon agentic tasks. This integration allows enterprise customers to access the model via fully managed APIs, including cross-Region inference and prompt caching, without managing their own infrastructure.
AWS Continuum, an autonomous code security tool, achieved an 89.0% end-to-end success rate on the CyberGym-E2E benchmark, identifying, demonstrating, and repairing vulnerabilities in real codebases. This result sets a new public standard, surpassing the previous high by 23.1 percentage points, and addresses the increasing volume of security vulnerabilities faced by security teams.
Anthropic's Claude Opus 5.5 and Sonnet 5.5 models are now available on Amazon Bedrock in AWS GovCloud (US) Regions. This allows organizations with strict regulatory and compliance requirements, including ITAR, to use AI-assisted development tools.
Amazon introduced Bedrock AgentCore Evaluations, a new capability within Amazon Bedrock AgentCore, to assess the performance of multi-agent systems. This tool helps enterprises ensure their AI agents are helpful, accurate, and explainable in real-world applications by measuring qualities like task success and instruction following.
AWS launched a public preview of Amazon Bedrock Managed Agents powered by OpenAI, integrating OpenAI's Agents API natively within AWS for building agents with existing AWS controls. Additionally, AWS expanded its model choices on Amazon Bedrock by adding OpenAI GPT-6.1 Sol, OpenAI GPT-6 Astra UltraFast mode, Anthropic Claude Sonnet 5.5, and SpaceXAI Grok 4.7.
A team of AWS developers open-sourced Pizza Bot, an application that provides an inbox-style interface for managing background AI agent tasks. This tool allows AI agents to perform scheduled or webhook-triggered work, delegate tasks, and pause for human approval, offering a way to manage long-running AI workflows without continuous monitoring.
Amazon Bedrock AgentCore now allows Claude Desktop users to integrate secure web search capabilities, addressing the limitation of Claude's knowledge cutoff. This integration uses AgentCore Gateway and a managed web index, keeping all query traffic within AWS infrastructure for enhanced security.
DigitalOcean introduced Managed Agents in public preview, providing a managed cloud infrastructure layer for AI agents. This offering aims to simplify the deployment and management of AI agents by handling runtime environments, tool access, and serverless AI inference, addressing challenges developers face with traditional VM setups.
This article details how to use Amazon S3 Vectors as a persistent memory layer within the NVIDIA NeMo Agent Toolkit (NAT), deployed on Amazon Elastic Kubernetes Service (Amazon EKS). It provides an implementation guide for building agent memory, moving from architectural concepts to practical application.
Amazon Bedrock AgentCore now supports ambient agents, which respond to event streams like S3 uploads or database changes, pausing for human input when necessary. This allows AI agents to automate tasks by reacting to infrastructure events rather than solely relying on chat-based user prompts, addressing limitations of traditional AI agent models.
Amazon Bedrock Knowledge Bases now allow users to query claims data using natural language. This feature helps policyholders, contact center agents, and adjusters quickly find and combine evidence from various unstructured documents like diary entries, repair estimates, and police reports.
AWS Bedrock AgentCore now offers Runtime Instances, a new compute option for persistent, long-running agent workflows. This enables multi-agent systems to collaborate over several days, share context, and utilize GPUs and persistent volumes, addressing limitations of serverless MicroVMs.
Amazon Bedrock now offers Anthropic's Claude Opus 5, Claude Sonnet 5, and Claude Haiku 4.5 models for in-country inferencing in India. This allows customers to process data within India's AWS Regions, addressing local data residency requirements.
Amazon Bedrock now supports Anthropic's Claude Opus 5 and Claude Sonnet 5 models for in-region inference in Seoul, and Claude Sonnet 5 in Singapore. This expansion allows organizations with strict data residency requirements in these regions, such as those in financial services or healthcare, to utilize these AI models at scale while keeping all data processing within the specified AWS Region.
Amazon Nova Act and Bedrock AgentCore can be used for agent-driven synthetic monitoring, which emulates user journeys to detect problems faster than traditional methods. This approach moves beyond brittle UI scripts to validate critical workflows continuously. The post provides an architecture and implementation example for this method.
NarrateAI, an AI assistant for business intelligence, has implemented five quality assurance techniques on Amazon Bedrock to achieve approximately 99% numerical accuracy in real-time LLM responses. These techniques address issues like hallucinated metrics, API throttling, and validation latency in production LLM applications. The implementation aims to provide reliable, data-driven insights for executives during live business reviews.
OpenAI launched its Agents API in public beta, and Cursor released Projects, both focusing on coordinating multiple AI agents for software development. These releases indicate a growing industry consensus on using a coordinator-worker architecture to manage AI agents, addressing limitations of single-agent systems.
A guide details how to construct a multi-account AI agent architecture using Amazon Bedrock AgentCore Gateway and Model Context Protocol (MCP). This setup allows AI agents to access data across multiple AWS accounts without centralizing or copying it, addressing enterprise needs for data isolation and independent deployment lifecycles.
Dutch retailer HEMA developed an internal AI assistant named HAL using Amazon Bedrock AgentCore and Model Context Protocol (MCP) to consolidate fragmented internal knowledge. This solution addresses the challenge of engineers navigating multiple disconnected portals to find information, providing instant answers across various roles and tools.
Amazon Bedrock now supports open weight models for AI coding agents, allowing developers to run these tools privately and cost-effectively within their AWS accounts. This setup addresses data residency, cost, and model flexibility concerns often associated with third-party AI coding solutions. The article details how to use OpenCode, an open-source terminal-native AI coding agent, with Bedrock's open weight models.
Amazon CloudWatch launched CloudWatch Omni, a new observability solution for generative AI and agentic workloads. This tool provides AI-powered monitoring, evaluation, and experimentation capabilities, addressing challenges with non-deterministic AI behavior and integrating directly into developer IDEs and a standalone web experience.
Strands Evals SDK and Amazon Bedrock AgentCore Evaluations have launched new evaluators designed to assess the performance of skill-equipped agents. These tools address specific failure modes in modular agent systems, such as incorrect skill selection or incomplete instruction following, which general output-quality metrics often miss.
Reactiv, a mobile commerce product for Shopify merchants, implemented Amazon Bedrock AgentCore to automate updates for native mobile apps. This integration reduced merchant configuration time by 80% and accelerated production deployment by 33% for their AI Scheduler feature. The AI Scheduler allows merchants to use natural language to schedule autonomous app updates, addressing the need for constant content freshness without manual intervention.
Trane Technologies implemented an AI-powered agentic solution on Amazon Bedrock AgentCore, reducing the time required for HVAC operational diagnostics from 20 minutes to 20 seconds. This improvement allows Trane to shift from reactive responses to proactive, data-driven optimization for its millions of connected HVAC assets worldwide.
AWS and Salesforce Agentforce now integrate to process unstructured public sector data like body camera footage and scanned documents. This integration uses Amazon Bedrock Data Automation and the Model Context Protocol (MCP) to convert raw evidence into structured insights, allowing Agentforce users to query processed data directly within Salesforce.
Benchling implemented a defense-in-depth security architecture using Amazon Bedrock AgentCore to run AI agent-generated scientific code for thousands of life sciences tenants. This architecture addresses the challenge of securely executing untrusted code at scale while preventing data exfiltration and ensuring per-tenant data isolation. The system processes over 600 code execution sessions daily across more than 250 tenants weekly without security incidents.
Moonshot AI's Kimi K3, an open-weight model with 2.8 trillion parameters, is now available on Amazon Bedrock, offering enhanced capabilities for coding and knowledge work. This integration provides users with a new option for AI development, featuring native vision, a 1-million-token context window, and improved scaling efficiency.
Amazon published a guide on migrating multi-model AI agents from self-managed infrastructure to Amazon Bedrock AgentCore runtime. This migration allows organizations to reduce infrastructure management overhead by offloading container orchestration, scaling, identity, and observability to the managed runtime.
Amazon announced an update to its Bedrock AgentCore runtime, which provides a managed compute layer for deploying and running AI agents. The new runtime offers better memory management, consistent cold start times, and more precise cost tracking, addressing the evolving needs of production AI agents.
AWS has made Gemma 4 open weight models generally available on Amazon Bedrock within the AWS European Sovereign Cloud. This allows European organizations to run generative AI workloads while maintaining data residency and operational controls within the EU, meeting digital sovereignty requirements.
Wood Mackenzie developed APEX (Agentic Platform for Energy eXperience), a shared platform built on Amazon Bedrock AgentCore, to address challenges in deploying AI agents to production. This platform centralizes orchestration, safety, observability, identity, and connectivity, allowing teams to focus on business logic rather than rebuilding infrastructure for each agent.
Amazon Bedrock AgentCore launched a system prompt optimizer to improve agent quality by analyzing production traces and proposing configuration changes. This feature uses AgentCore Observability data and a reward signal to generate revised system prompts, aiming to automate the optimization process for AI agents.
Amazon Web Services (AWS) published a guide on building a serverless pipeline for redacting Personally Identifiable Information (PII) from documents using Amazon Bedrock Data Automation. This solution addresses the challenges of manual PII redaction and limitations of traditional OCR and ML models by leveraging generative AI for document understanding.
AWS released a new Step Functions pattern that integrates AI agents into airline rebooking processes, ensuring deterministic code controls reservation changes and payments. This pattern allows Amazon Bedrock AgentCore agents to suggest itineraries and compensation messages, which are then validated by code before any actions are taken. The development matters because it provides a framework for safely deploying AI agents in critical business operations by separating agent proposals from direct system actions.
Google has made Agent Substrate available on Google Kubernetes Engine (GKE), providing an open-source runtime for executing AI agents with higher density and improved security. This aims to address challenges in scaling autonomous agents, such as opaque trust boundaries, tool access friction, massive bursts, and idle compute, by offering a specialized infrastructure.
Abnormal AI, a behavioral security service, has deployed Amazon Bedrock AgentCore Code Interpreter to power its real-time inline email threat detection systems. This integration allows Abnormal AI's agents to execute code dynamically at scale, processing billions of messages to identify and block threats before they reach inboxes.
Amazon Bedrock AgentCore now includes a Consent portal, a managed web experience and session binding endpoint for AgentCore Gateway, simplifying OAuth consent for AI agents accessing third-party services. This update removes the need for customers to build custom infrastructure for session binding, making it easier to integrate AI agents with services like GitHub and Slack while maintaining user authentication.
Ninth Wave developed Compass, an AI-enabled onboarding assistant powered by Amazon Bedrock AgentCore, to streamline the integration of financial institutions into open finance networks. This tool addresses the challenge of varying API standards among banks by normalizing data to FDX standards, reducing the time required for API validation and field mapping.
AWS published a guide detailing an 8-step decision framework for customizing generative AI models, ranging from prompt engineering to training custom models. This framework helps users select the appropriate customization approach for their specific workload on AWS, avoiding over-engineering or under-investment.
AWS has introduced new monitoring capabilities for multi-agent systems in production environments. These tools, AWS DevOps Agent and Amazon Bedrock AgentCore Evaluations, address the limitations of traditional infrastructure monitoring in detecting agent-specific issues like incorrect responses or routing problems.
Amazon Bedrock AgentCore now supports building interactive Model Context Protocol (MCP) applications with HTML widgets for AI hosts like ChatGPT and Claude. This allows organizations to integrate their services with rich user interfaces into various AI platforms without coupling to a single host. The integration enables interactive experiences beyond plain text within AI applications.
AWS open-sourced Pizza Bot, an application providing an email-like inbox for managing background AI agents. This tool addresses the challenge of monitoring AI agents that operate autonomously over extended periods, allowing users to review completed tasks and human-decision requirements asynchronously.
OpenAI has released the Agents API, providing applications with access to the Codex harness for AI agent orchestration. This API manages sessions, context, and recovery, allowing applications to define tools and execution environments for agents that can operate in sandboxes.
AWS released updates to Amazon Bedrock, introduced AgentCore, and open-sourced the Strands Agent Harness SDK in August 2026. These updates aim to provide tools for building and scaling AI applications, particularly focusing on agentic applications and enabling models to handle more complex tasks with greater context.
Heurist Finance developed an AI-powered investment workbench for retail investors on Amazon Bedrock AgentCore. This platform enables Heurist to integrate institutional-style financial workflows, including market data analysis, portfolio stress-testing, and position monitoring, while managing paid data access per query.
A guide details how to implement automated quality gates for AI agents deployed on Amazon Bedrock AgentCore using GitHub Actions. This setup evaluates agent performance and blocks pull requests if code changes cause regressions, ensuring continuous quality control for AI applications.
HPE Zerto built an agentic troubleshooting system powered by Amazon Bedrock to assist with cyber resilience, disaster recovery, and continuous data protection. This system aims to provide natural language insights for faster recovery decisions in hybrid and multi-cloud environments. It addresses challenges like fragmented information and the need for quick, actionable guidance during outages.
DiDi International Business Group partnered with AWS to develop an intelligent contact center quality assurance (QA) system using Amazon Bedrock, replacing a third-party solution. This new system improved intent verification accuracy from 38% to 86% and reduced manual summarization time for Voice of Customer analysis from hours to minutes. The implementation provides DiDi with a transparent and self-owned AI architecture for its customer experience operations across multiple business lines and languages.
AWS published instructions for deploying a multimodal WhatsApp ordering assistant using Amazon Bedrock AgentCore and Amazon Nova 2. This solution allows quick-service restaurants to consolidate customer interactions across text, voice notes, and calls into a single WhatsApp channel, maintaining customer history across communication methods. The deployment guide provides a method for businesses to integrate AI-powered ordering directly into a widely used messaging platform.
Amazon Bedrock AgentCore now supports memory lifecycle policies to manage what long-running AI agents remember and forget. This feature helps prevent agents from accumulating outdated context, which can degrade response quality and create compliance risks.
Intuit developed EWOK Agent, an AI-powered assistant built with Amazon Bedrock, to enhance its existing disaster recovery system, EWOK. This agent addresses decision-making challenges during failovers, which previously relied on experienced engineers, by providing AI-driven reasoning and handling exceptions.
Reference implementations demonstrate using Amazon Bedrock AgentCore for AI-driven development lifecycle patterns, specifically for generating Mermaid ER diagrams from SQL schemas and automated code security analysis. These implementations aim to bridge the gap between conceptual AI frameworks and practical application in software development. This matters to developers looking for practical examples of integrating AI agents into their development workflows.
Amazon Bedrock AgentCore now supports migrating existing agentic workloads, such as LangGraph customer support agents, to its managed environment. This migration offloads operational burdens like session management, state persistence, and tool authentication from developers, allowing them to focus on agent reasoning.
A guide details how to deploy a customer-operated LiteLLM gateway on Amazon ECS, connecting it to an OpenAI model on Amazon Bedrock, and configuring Codex to use the gateway's Responses API. This setup provides centralized enterprise controls for generative AI coding agents, enabling consistent management of model access, consumption, budgets, and rate limits.
An AWS team developed an automated solution using Amazon Bedrock to detect silent content failures in dashboards hosted on AWS Insights, reducing detection time from up to 72 hours to less than 1 hour. This system visually analyzes dashboards with large language models to identify blank, stale, or incorrect data before users encounter issues.
Amazon Bedrock AgentCore is used by a global interdealer broker to automate architecture documentation, integrating with existing CI/CD workflows. This solution analyzes codebases, generates diagrams, and maintains searchable documentation, addressing challenges like manual processes and rapid obsolescence.
Jamf developed a system to enforce real-time spending limits for individual engineers using Amazon Bedrock, addressing the challenge of unpredictable generative AI costs. This solution tracks daily AI spend per user and applies tiered model restrictions, such as denying access to premium models at budget thresholds, to manage costs without interrupting work sessions.
t54 has developed a trust layer for autonomous agent payments on Amazon Bedrock AgentCore, enabling agentic systems to make micropayments for third-party services without human approval. This system has processed over 20 million transactions, addressing the challenge of governing agent spending at scale while maintaining security and compliance. The solution combines Amazon Bedrock AgentCore payments for infrastructure with t54's x402-secure for real-time endpoint scoring.
AWS published a guide detailing how to integrate an AgentCore Runtime hosted Model Context Protocol (MCP) server with Amazon Quick. This integration allows AI agents in Amazon Quick to access external data and tools, supporting standardized and secure interactions with files, databases, and APIs. The process aims to promote reusability of AI tools and agents across different client applications.
The Agentic Resource Discovery (ARD) open specification has been introduced to enable AI agents to find and connect to tools across diverse cloud and platform environments. This addresses a limitation of the Model Context Protocol (MCP), which assumes agents already know the specific server they need to connect to, making it difficult for organizations with distributed resources.
AWS has made its Agent Registry generally available, providing a centralized, searchable, and governed catalog for AI agents, tools, skills, and custom resources. This service addresses challenges in discovering and governing agentic AI systems at scale, such as lack of inventory, cross-team discovery, and governance.
This post details how to implement an enterprise agentic retrieval solution using Amazon Bedrock Managed Knowledge Bases and AWS CloudFormation, focusing on observability for multi-knowledge base RAG workflows. The solution allows an agent to reason, route questions across multiple knowledge bases, and synthesize cited answers, addressing the operational challenges of complex RAG systems.
Amazon Bedrock Managed Knowledge Base now supports building multi-tenant agentic chat applications on enterprise data. This service handles ingestion, storage, embedding, and ranking, along with built-in agentic retrieval that enforces tenant isolation and access permissions. This allows teams to deploy document chat features without building extensive retrieval infrastructure or isolation logic.
AWS has open-sourced Kiro Crew, a system that allows developers to assign asynchronous coding tasks to AI agents for work like incident investigation and migrations. This release enables persistent, multi-session development work with features such as shared memory and scheduled jobs, making AI agents more autonomous in development workflows.
Amazon Bedrock now supports OpenAI's GPT-5.6 Terra and Luna models for in-country inferencing in India, allowing organizations with local data processing requirements to use these models at scale. This update enables applications to process text and image inputs within India, addressing data residency needs for sectors like financial services, healthcare, and public sector.
Amazon Bedrock Guardrails now extends protection beyond the model boundary to include tool interactions through the Strands Agents SDK. This allows validation of data flowing to and from external tools and systems, addressing gaps where model-level guardrails previously did not apply. The extension helps prevent issues like unchecked PII in tool parameters or policy-violating content in external data sources.
Google Cloud has launched Cloud Run instances, a new offering designed for deploying long-lived, stateful workloads like personal AI agents. This service provides dedicated, singleton compute runtimes with continuous operation capabilities, addressing the limitations of existing Cloud Run services for such applications.
Amazon Bedrock AgentCore Evaluations now allows evaluation of any agent framework by using OpenTelemetry for telemetry data. This update addresses the fragmentation in AI agent development by decoupling evaluation from specific SDKs or LLM clients, enabling teams to use their preferred frameworks while still utilizing Bedrock's evaluation services.
Natera, a diagnostics company, developed an automated voice agent for patient appointment scheduling using Amazon Bedrock AgentCore. This system replaces manual scheduling calls, allowing patients to book phlebotomy appointments conversationally.
Amazon Bedrock AgentCore can now connect to knowledge bases hosted in separate AWS accounts, specifically those backed by Amazon Redshift Serverless. This enables organizations to maintain workload boundaries and avoid data copying while generating natural-language answers from structured data.
Amazon OpenSearch Service now includes MCP Apps, which integrate interactive observability visualizations directly into AI assistant chat windows. This change allows users to verify agent-generated root cause hypotheses without switching applications, addressing a bottleneck in agentic observability workflows.
AWS introduced Agent Registry, a centralized catalog for AI agents, Model Context Protocol (MCP) servers, and tools, alongside the open Agentic Resource Discovery (ARD) specification. This aims to address the challenge of discovering and managing AI resources across different environments as organizations scale their AI agent use.
AWS released aws-bench, an open-source benchmark designed to evaluate AI agents' accuracy in completing real AWS tasks using live cloud environments. This benchmark provides a more accurate assessment of agent performance on developer-relevant tasks compared to static benchmarks, addressing growing skepticism about agent evaluation methods.
Amazon introduced Bedrock AgentCore Gateway, a new capability within Amazon Bedrock AgentCore, to provide a single, secure entry point for AI agents accessing organizational tools. This release addresses security risks associated with ungoverned AI agent access, such as credential sprawl and audit gaps, by centralizing authentication, authorization, and policy enforcement.
Amazon Bedrock now supports cross-Region inference for OpenAI GPT-5.6 models, including Sol, Terra, and Luna variants, across more than 25 AWS Regions. This feature uses inference profiles to route requests to available compute capacity, improving throughput and consistent performance under load by drawing from a broader pool of resources.
Amazon Bedrock AgentCore now includes an AI-driven Policy Authoring tool that converts natural language policy specifications into Dogwood formal specifications. This expansion allows users to generate policies enforcing temporal and trajectory constraints, integrate with Guardrails for content detection, and restrict tool input parameters, enhancing control over AI agent actions.
AWS Professional Services developed a multi-agent framework running on Amazon Bedrock AgentCore to automate and scale cloud migrations. This framework reduced infrastructure as code development time from weeks to minutes for over 300 applications, addressing common bottlenecks in large-scale enterprise migrations.
AWS published a guide on implementing intelligent security for Fast Healthcare Interoperability Resources (FHIR) APIs using Amazon Bedrock. This solution aims to enhance security monitoring, automate data sensitivity classification, and streamline compliance reporting for healthcare data.
Amazon Bedrock AgentCore's Web Search feature now includes runtime domain and publish date filtering, allowing AI agents to control source material and freshness. This update also expands Web Search availability to the eu-west-1 (Dublin) and ap-northeast-1 (Tokyo) AWS Regions, reducing latency and addressing data proximity requirements for customers in those areas.
AWS introduced asynchronous invocation patterns for Amazon Bedrock AgentCore agents in serverless pipelines to reduce idle compute costs. These patterns address the latency of AI agent processing by releasing the caller's compute during wait times, preventing billing for inactive resources.
Amazon Bedrock AgentCore now supports patterns for propagating user authorization context in AI agents, ensuring agents only access data users are authorized to see. This approach enforces least privilege access without requiring authorization logic within the agent's code, addressing a key security risk in AI agent deployments.
Amazon Web Services introduced runtime instances for Bedrock AgentCore, providing a new compute option that runs AI agents on managed Amazon EC2 instances for up to fourteen days. This update allows for multi-agent collaboration with shared file systems and GPU acceleration, addressing limitations of the previous eight-hour microVM ceiling.
AWS published guidance on implementing custom authentication, specifically HTTP Basic Authentication, for tools integrated with Amazon Bedrock AgentCore Gateway using a request Lambda interceptor. This allows organizations with legacy systems to integrate AI agents without immediately modernizing their authentication mechanisms, while keeping credentials isolated from the agent.
Amazon Bedrock AgentCore Payments has reached general availability, allowing AI agents to autonomously pay for APIs and content. This service integrates with Coinbase and Stripe Privy wallets, enabling secure microtransactions for production AI workloads.
Amazon Bedrock now supports vector-prompt document classification through a multi-agent solution, which helps classify documents like policies and regulatory forms. This approach combines textual reasoning with visual pattern recognition to improve accuracy in complex classification tasks.
Axonius, an asset intelligence platform, has adopted Amazon Bedrock AgentCore to build secure, multi-tenant AI agents for its SaaS offerings. This implementation addresses common challenges for Independent Software Vendors (ISVs) in deploying AI agents, such as security, scalability, and cost tracking, particularly for managing individual customer environments.
AWS and the OpenClaw Foundation collaborated on a guide demonstrating how to integrate OpenClaw agents with Amazon Bedrock AgentCore payments. This integration allows autonomous agents to handle programmatic payments for services requiring an HTTP 402 response, enabling continuous operation without human intervention for each transaction. The capability addresses the need for agents to interact with pay-per-use APIs and content services, particularly for microtransactions where traditional card processing is inefficient.
AWS open-sourced Dogwood, a policy language that extends Cedar to govern sequences of agent tool calls by allowing rules to consider past actions. This development enables more complex control over AI agent workflows, addressing limitations of Cedar which only evaluates single requests.
AWS published a guide on integrating OpenAI-compatible endpoints on Amazon SageMaker AI with Amazon Bedrock AgentCore runtime to build agentic workflows. This integration allows specialized agents to use different models, including custom or cost-optimized ones on SageMaker, alongside managed models on Bedrock, within a single framework. The approach provides flexibility for model choice, cost optimization, and data residency for AI applications.
AWS published guidance on configuring AgentCore Observability for AI agents deployed outside of AWS environments, including on-premises or other cloud providers. This allows users to centralize monitoring for AI agents built with frameworks like Strands Agents, LangGraph, and CrewAI, regardless of their deployment location.
Amazon Bedrock AgentCore Browser Tool, combined with Strands Agents, provides a fully managed browser service for AI agents to interact with legacy web applications. This tool addresses the challenge of automating older systems that lack modern APIs and require human-like interaction, reducing manual data entry and errors in critical business operations.
Amazon Bedrock AgentCore can be used to build multi-agent systems for M&A due diligence, automating data gathering, analysis, and compliance checks. This approach addresses challenges like slow cycles, fragmented data, duplicated effort, and governance requirements in traditional M&A processes.
OneAdvanced, a UK enterprise software provider, implemented over 50 AI agents using self-hosted Llama 4 Maverick and Llama Guard 4 models on AWS to meet UK data sovereignty requirements. This approach allowed them to maintain data residency within the UK, which is critical for their customers in regulated industries.
Solv Labs developed an AI agent-payments workflow using Amazon Bedrock AgentCore payments, integrating its ORACLE policy engine and ICME PreFlight for compliance verification. This system enables auditable and governed transactions for AI agents, addressing the need for proof and accountability in autonomous financial operations. The solution allows enterprises to manage agent spending in regulated environments with full audit trails and pre-authorization.
OpenAI's Daybreak Red and Daybreak Blue models, designed for cybersecurity defense, are now available to eligible customers on Amazon Bedrock. This integration allows cyber defenders to use advanced AI for vulnerability analysis and remediation within AWS's controlled infrastructure.
Anthropic released a reference deployment guide for its Claude apps gateway on AWS, providing a self-hosted governance layer for enterprise use of Claude Code and Claude Desktop. This deployment offers centralized controls for authentication, model access, cost attribution, and spend enforcement for AI administrators.
nOps, a cloud optimization solution, transitioned its FinOps analytics to Amazon Bedrock AgentCore, reducing the time to ship FinOps agents by 75%. This change improved product delivery, response quality, and reduced operational complexity for managing cloud commitment optimization across AWS, GCP, and Azure.
Cohere Health developed Cohere Policy Studio using Amazon Bedrock AgentCore to convert static clinical policies into structured, machine-readable data. This aims to automate prior authorization processes in healthcare, which are currently manual and vary widely.
The TReNDS Center at Georgia State University implemented an automated root-cause analysis system using Amazon Bedrock, AWS Lambda, and other AWS services. This system detects errors in real time, enriches them with log context and source code, and provides AI-powered analysis to reduce manual investigation time.
Microsoft has launched a dedicated AI Gateway tier for Azure API Management in public preview, designed to govern AI models and MCP tools rather than traditional APIs. This new tier centralizes the management of multiple AI providers and models, offering features like token limits, quotas, and cost governance at the gateway level. It matters because it provides a specialized solution for managing the growing complexity of AI model deployments across various providers, enabling better control and cost management for platform and application teams.
AWS has introduced Runtime Instances, a new compute option within Amazon Bedrock AgentCore Runtime, providing persistent, managed infrastructure for complex AI agent workloads. This allows AI agents to run continuously for multiple days, access GPUs, and coordinate within shared sessions, addressing infrastructure challenges for production AI deployments.
AWS introduced Dogwood, an open-source policy language and reference interpreter designed to govern sequences of AI agent tool calls, rather than individual actions. This development allows for more complex policy enforcement, considering prior events and overall workflow, which is crucial for the safe and controlled operation of AI agents.
Amazon Bedrock Guardrails intervention data can now be routed to Amazon Security Lake, allowing security teams to analyze AI-related incidents alongside other security telemetry. This integration enables unified querying of guardrail events with identity, network, and application security data for comprehensive incident investigation.
Amazon Bedrock AgentCore now includes temporal policies, which allow for stateful authorization rules that evaluate AI agent requests based on their session history. This update addresses security challenges in AI agents by preventing issues like incorrect tool sequencing or data fabrication, which stateless policies cannot catch.
Amazon Bedrock's AgentCore gateway now supports rate limiting for AI traffic, allowing per-user control over requests, concurrent connections, and token throughput. This feature helps maintain the availability of downstream services by managing traffic spikes to tools, inference models, and agents.
AWS Kiro has replaced its multiple agent harnesses with a single architecture built around the Agent Client Protocol (ACP), allowing Kiro's clients to communicate with its agent through a shared protocol. This change enables developers to choose their coding tools and AI agents independently, standardizing the client-agent interface rather than using proprietary implementations.
AWS released guidance on integrating Codex with Amazon Bedrock, OpenTelemetry, and Amazon CloudWatch to provide visibility into coding agent usage. This integration allows organizations to monitor adoption, consumption, and reliability of Codex activity through an AWS-native view, without adding a centralized proxy.
AWS published a guide on how to configure Claude Code on Amazon Bedrock to enforce single-Region data residency, specifically for compliance requirements. This allows organizations to ensure that all prompts, completions, and intermediate processing remain within a designated AWS Region, addressing strict data residency mandates.
This article details how to use Agent Skills to manage the lifecycle of Amazon Bedrock Automated Reasoning policies from a coding agent. Agent Skills provide specialized knowledge and workflows to coding agents, enabling them to build, test, deploy, and validate policies more effectively. This approach allows for repeatable, reviewable, and code-driven policy management.
PDI Technologies created PDI Brew, a tool that allows non-technical employees to deploy multi-tenant web applications on AWS using natural language descriptions. This system utilizes Amazon Bedrock and AWS Lambda to automate the provisioning of AWS resources and integrate AI capabilities into the deployed applications.
LendingTree has developed a multi-agent AI mortgage assistant on Amazon Bedrock to help borrowers navigate the home-buying process. This assistant educates users, understands their financial situation, and provides tailored mortgage options through natural conversation, addressing the complexity of mortgage decisions.
Mobileye implemented an AI Support Agent using Amazon Bedrock AgentCore to automate routine internal ticket status inquiries, which previously required manual steps across multiple systems. This solution reduced response times by 90% and achieved over 95% accuracy, freeing skilled engineers from repetitive tasks and improving efficiency.
Amazon Bedrock AgentCore harness is now generally available, providing scaffolding for production AI agents with persistent memory and tools. A new open-source n8n community node integrates this capability, allowing users to build agents in n8n's visual editor without writing infrastructure code.
Amazon has made Web Search generally available on Amazon Bedrock, allowing foundation models to access current web knowledge for grounding responses. This integration removes the need for developers to integrate and maintain third-party web search providers, simplifying the process of grounding models and reducing hallucinations.
Amazon Web Services published a tutorial on deploying an automated web insight extraction solution using Amazon Bedrock AgentCore Browser, Amazon Bedrock for AI analysis, Amazon OpenSearch Serverless, and AWS Lambda. This solution addresses the challenge of manually extracting insights from numerous websites, especially those with dynamic content, by providing a resilient, AI-powered pipeline.
Amazon Bedrock now offers automatic policy refinement for its Automated Reasoning feature, automating the diagnosis and fixing of policy failures. This update addresses a significant friction point in policy development by proposing formal-logic fixes for failing tests, improving the efficiency of creating and validating policies.
Amazon Bedrock AgentCore Observability and Amazon CloudWatch can be used to identify and diagnose performance bottlenecks and memory issues in AI agents running in production. This helps maintain user trust and control costs by addressing slow response times and unbounded memory growth before they impact users.
Amazon Bedrock has launched Advanced Prompt Optimization, a new tool designed to automate the process of migrating and optimizing prompts across up to five different generative AI models. This feature addresses the manual effort and time previously required for prompt engineering, which often led to model lock-in and underperformance.
Google Cloud announced enhancements to its borderless Lakehouse, built on Apache Iceberg, enabling AI agents to access and act on data across AWS Glue, Databricks Unity, and Snowflake Horizon. This update allows for zero-copy, cross-cloud analytics and bidirectional interoperability, which matters because it reduces data movement costs and complexity for AI-driven workflows.
Microsoft introduced a reference architecture for routing large language model (LLM) agent traffic on Azure Kubernetes Service (AKS). This architecture optimizes cost and performance for agentic workloads by intelligently directing LLM calls to appropriate models and GPU resources.
Amazon published best practices for configuring Bedrock Guardrails in code generation workflows to prevent issues like throttling, increased costs, and latency. These guidelines address the unique throughput characteristics of AI-powered coding assistants, which generate long streaming outputs and handle concurrent developer sessions. Proper configuration helps detect and block unsafe code patterns and sensitive information while maintaining efficiency.
Motorway and AWS Prototyping and AI Customer Engineering (PACE) developed an evaluation pipeline for AI agents, reducing incorrect results from 1 in 8 queries to 1 in 50. This pipeline combines the Strands Agents SDK with Amazon Bedrock AgentCore to improve reliability and detect issues faster in production AI agents.
Amazon Bedrock AgentCore optimization now provides insights to detect and prioritize behavioral failures in deployed AI agents, including those that do not generate error signals. This feature helps identify patterns affecting agent performance at scale, moving beyond individual trace inspection.
Amazon Bedrock Managed Knowledge Base now includes agentic retrieval capabilities, allowing it to handle multi-part, comparative, and exploratory questions more effectively. This enhancement addresses limitations of single-shot retrieval, which often fails to provide comprehensive answers for complex queries spanning diverse document types.
Amazon, Microsoft, and Google have developed enterprise agent platforms with a shared architectural framework. Their platforms—Amazon Bedrock AgentCore, Microsoft Foundry, and Gemini Enterprise Agent—feature core components like runtime and observability, signaling a shift from fragmented libraries to unified production agents.
Amazon, Microsoft, and Google have introduced enterprise agent platforms that share a common architecture across their offerings. This shift indicates a move towards unifying the agent ecosystem, enhancing scalability and deployment for enterprises, similar to past advancements in platform as a service.
Couchbase has implemented Amazon Bedrock to support its AI-powered Capella iQ, enabling flexibility in using multiple foundation models. This architecture enhances operational resilience and supports high availability across AWS Regions, accommodating increased enterprise usage.
AWS has launched Loom, an open-source platform for building and governing AI agents, featuring built-in security controls. It aims to provide organizations with a guideline for deploying AI agents at scale on AWS while addressing common operational challenges.
Amazon Bedrock has launched its Managed Knowledge Base, a streamlined solution for building enterprise search capabilities. This service simplifies the setup process for knowledge retrieval systems by integrating data ingestion, search infrastructure, and access control, eliminating the complexity of traditional setups.
The article details the development of a voice ordering system for restaurants using Amazon Bedrock AgentCore and Nova 2 Sonic. It addresses the issue of missed customer calls during peak hours, demonstrating how to create a telephony AI that can handle orders and bookings autonomously, thus improving operational efficiency.
Amazon has integrated computer vision, Strands Agents, and Model Context Protocol (MCP) for AI applications. This streamlined approach facilitates the processing of visual information and decision-making, enhancing accessibility for developers.
Thrad.ai implemented a multi-agent system using Strands Agents and Amazon Bedrock for automating social intelligence. This system enhances prospect discovery and personalized outreach, significantly improving efficiency for lead generation and sales outreach.
Amazon Nova Act's QA Studio now includes features for batch regression testing and CI/CD integration. This improves software delivery by allowing concurrent execution of multiple test cases, thereby streamlining the quality assurance process.
Bluesight has developed Prism, an agentic AI solution using Amazon Bedrock to enhance healthcare compliance across its product suite. This AI integration aims to streamline the complex compliance processes, reducing manual labor for hospitals managing drug pricing program compliance.
Amazon Bedrock AgentCore Gateway now supports OAuth 2.0 Token Exchange for multi-tenant AI agents, allowing better identity handling. This implementation helps maintain an audit trail by securely propagating user identity across different downstream APIs without compromising security.
A guide outlines how to create a semantic layer on AWS that integrates Stardog with Amazon Bedrock AgentCore for improved enterprise analytics. This solution enables generative AI agents to autonomously query disparate data sources without the need for ETL, streamlining data access and analytics.
KTern.AI has created AI agents designed for SAP transformations using Amazon Bedrock AgentCore, enabling autonomous orchestration of complex workflows. This shift aims to improve efficiency in enterprise-scale SAP projects, offering automation without the need for custom infrastructure.
Amazon Bedrock AgentCore and Mistral AI Studio simplify the development of a production-ready ecommerce MCP server. This integration reduces custom API work and security risks, allowing faster deployment of AI-driven customer experiences within ecommerce platforms.
Amazon Bedrock AgentCore can now leverage AWS WAF for enhanced security via two architecture patterns, addressing authentication issues during health checks. This integration enhances protection against web threats while managing authenticated traffic among existing AWS services.
Jamf has launched AI Governance to manage AI applications like Claude and OpenAI Codex on Macs using Amazon Bedrock. This development allows organizations to configure and control AI applications centrally, enhancing security and compliance across devices.
Amazon Bedrock has enhanced its platform with features to enforce zero data retention policies, allowing organizations to manage data handling after inference requests. This change is significant as it provides users with means to ensure compliance and control over how prompts and outputs are retained, particularly when sharing data with third-party models.
The tutorial outlines the creation of a serverless image editing agent using Amazon Bedrock's AgentCore harness. This simplifies the implementation by automating orchestration and tool management, allowing users to edit images via natural language requests with minimal custom code.
Amazon Bedrock AgentCore enables developers to build an AI support companion for AWS management. This solution reduces investigation time by consolidating multiple support tasks into a single conversational interface, streamlining operations for AWS support teams.
Amazon Bedrock introduces MiniMax models, offering three open-weight foundation models tailored for software engineering and AI tasks. These models ensure that organizations can maintain security and compliance while leveraging advanced AI capabilities.
Amazon Bedrock in AWS GovCloud (US) now supports NVIDIA Nemotron and OpenAI open-weight GPT OSS models. This enables U.S. government agencies to utilize advanced AI capabilities while ensuring compliance and data security within the requisite isolation boundaries.
Amazon has launched the Model Profiler, an open source tool that simplifies model selection in Amazon Bedrock. It aggregates model metadata from various sources into a unified interface, enabling teams to efficiently compare models' capabilities, pricing, and availability.
AWS is introducing new advanced AI models, including Anthropic’s Claude Fable 5, on its Bedrock platform while emphasizing security. The updates aim to provide customers with faster access to cutting-edge models without compromising safety and privacy.
Amazon has launched the AG-UI protocol to facilitate dynamic user interactions with AI agents on Bedrock AgentCore. This open protocol enhances the integration between various agent frameworks and frontend libraries, allowing for more sophisticated interactions in AI applications.
Amazon Bedrock now supports managed entitlements, allowing centralized subscription management for AI models across multiple AWS accounts. This feature reduces operational overhead for organizations using third-party models by eliminating the need for individual AWS Marketplace permissions in workload accounts.
Amazon Bedrock AgentCore now supports resource-based policies, enabling SaaS providers to centralize access control for multi-tenant AI applications. This feature allows distinct security configurations for different tenants within a shared infrastructure, improving compliance and management.