AWS Machine Learning

Amazon Quick and Amazon Bedrock enhance Retrieval Augmented Generation (RAG) by implementing real-time access control lists (ACLs) to ensure AI-generated insights respect sensitive permissions from sources like SharePoint and Google Drive. This dual-layer approach combines pre-retrieval filtering with real-time verification, addressing common security gaps and ensuring always-current permissions for enterprise users.

The introduction of real-time access control lists (ACLs) in Amazon Quick and Amazon Bedrock for Retrieval Augmented Generation (RAG) significantly enhances data security by ensuring that AI-generated insights comply with sensitive permissions. This development is crucial for builders and PMs as it mitigates security risks in enterprise applications, while investors should note its potential to increase adoption in sensitive sectors.

Qlik built Qlik Answers on Amazon Bedrock to provide grounded AI solutions for over 40,000 customers, enabling quick, sourced answers to complex enterprise queries. The system has already facilitated over 100,000 discoveries and improved response times by up to 75% for clients like Lintech International.

Qlik's development of Qlik Answers using Amazon Bedrock demonstrates how enterprise-scale AI can enhance data accessibility and decision-making efficiency, achieving a 75% improvement in response times. This signals to builders and PMs the viability of integrating AI into existing systems, while investors should note the potential for scalable solutions in the data analytics market.

AWS's six-week program successfully bridged the AI knowledge-capability gap for non-engineering professionals, enabling them to build a financial advisory tool, WealthWise, using Amazon Bedrock and Strands Agents SDK. Participants reported enhanced learning and confidence through structured mentorship and hands-on experience.

AWS's six-week program effectively equips non-engineering professionals with the skills to build AI tools like WealthWise using Amazon Bedrock, highlighting a growing trend in democratizing AI development. This signals an opportunity for builders and PMs to leverage accessible AI education, while investors may find potential in startups that can capitalize on this emerging talent pool.

Cornerstone OnDemand reduced database diagnosis time by 78% from 45 minutes to 10 minutes using Orion AI, powered by Amazon Bedrock. The multi-agent system streamlines workflows, automating manual tasks and enhancing operational efficiency for their Enterprise DataOps team.

Cornerstone OnDemand's use of Amazon Bedrock to reduce database diagnosis time by 78% demonstrates the tangible benefits of AI in enhancing operational efficiency. Builders and PMs can leverage similar AI solutions to streamline workflows, while investors should note the potential for significant cost savings and productivity gains in enterprise operations.

AWS introduces a context-aware AI assistant built on AgentCore and OpenClaw, enabling continuous interactions by retaining user context. This system, exemplified by a gardening assistant named Sprout, utilizes Amazon Bedrock's Claude models for text and image processing, costing approximately $1–2 per month for light use.

AWS's introduction of a context-aware AI assistant using AgentCore and OpenClaw represents a significant shift towards more interactive and personalized user experiences. For builders and PMs, this means opportunities to create applications that leverage continuous context retention, while investors should note the potential for scalable, subscription-based AI services that can drive recurring revenue.

AWS supports organizations in aligning with ISO/IEC 42005:2025 for responsible AI governance, emphasizing the importance of AI system impact assessments to manage risks effectively. The framework aids in integrating AI assessments into existing governance, enhancing enterprise-wide risk management and compliance.

AWS's support for ISO/IEC 42005:2025 in responsible AI governance signals a shift toward standardized risk management practices in AI development. Builders and PMs should integrate these frameworks to enhance compliance and mitigate risks, while investors may view adherence to such standards as a marker of a company's commitment to sustainable and responsible AI practices.

Amazon SageMaker HyperPod enables ML teams to share accelerated compute resources, necessitating robust governance for access, capacity, and workload management. The integration with SageMaker Unified Studio enhances project collaboration while maintaining essential administrative controls across organization, project, cluster, and workload layers.

The introduction of Amazon SageMaker HyperPod enhances resource sharing for ML teams, which can significantly improve project efficiency and collaboration. For builders and PMs, this means streamlined governance and workload management, while investors should note the potential for increased productivity and faster time-to-market for AI solutions.

Amazon SageMaker now allows users to manage HyperPod Spaces directly from SageMaker Studio, streamlining the process for data scientists and ML engineers. This new feature eliminates the need for command-line tools, enabling quick access to JupyterLab and Code Editor environments with just a few clicks, enhancing productivity.

The integration of HyperPod Spaces management directly within SageMaker Studio simplifies the workflow for data scientists and ML engineers, reducing the reliance on command-line tools. This enhancement can lead to increased productivity and faster iteration cycles, making it a significant development for teams focused on machine learning projects.

GLM 5.3 from Z.ai is now available on Amazon Bedrock, featuring 753 billion parameters optimized for coding and agentic tasks, with notable improvements in cybersecurity performance, achieving an 84.5 score on the CyberGym benchmark. The model supports cross-Region inference and prompt caching, allowing enterprise customers to reduce costs and latency without managing infrastructure.

The introduction of GLM 5.3 on Amazon Bedrock, with its 753 billion parameters and enhanced cybersecurity capabilities, enables builders and PMs to leverage advanced AI for coding and security tasks while optimizing costs and performance. Investors should note the model's cross-Region inference and prompt caching features, which can significantly improve enterprise operational efficiency and reduce infrastructure management burdens.

Anthropic's Claude Opus 5.5 and Claude Sonnet 5.5 are now available in AWS GovCloud, enabling AI-assisted development for regulated workloads. With FedRAMP Class D certification, these models support compliance with ITAR and DoD standards, while Claude Code automates coding tasks across various environments.

Anthropic's Claude Opus 5.5 and Claude Sonnet 5.5 are now available in AWS GovCloud, providing AI tools that comply with ITAR and DoD standards for regulated workloads. This development enables builders and PMs to leverage AI for faster, compliant coding, while investors can recognize the potential for growth in sectors requiring stringent regulatory adherence.

Amazon SageMaker introduces the aws-ai-ml skill for coding agents, enabling inference optimization and benchmarking. This skill allows agents like Kiro and Claude Code to generate executable SageMaker Python SDK v3 code, enhancing deployment configurations and performance comparisons, thus accelerating the transition from model to production.

Amazon SageMaker's introduction of the aws-ai-ml skill for coding agents allows developers to optimize and benchmark generative AI inference, streamlining the transition from model development to production. This enhancement not only improves deployment configurations but also increases the efficiency of coding agents like Kiro and Claude Code, making it a significant advancement for builders and PMs focused on AI integration.

Amazon Bedrock's agentic retrieval enhances LangChain's RAG applications by breaking complex queries into sub-queries, optimizing search accuracy and relevance. This method improves the retrieval process by ensuring sufficient evidence is gathered before finalizing responses, thus offering a more comprehensive answer to multi-part questions.

The integration of Amazon Bedrock's agentic retrieval with LangChain significantly enhances the accuracy and relevance of responses in retrieval-augmented generation (RAG) applications. For builders and PMs, this means improved user experience and efficiency in handling complex queries, while investors should note the potential for increased adoption of advanced AI solutions in various industries.

Downgrading user roles in Amazon Quick is essential for maintaining security and cost efficiency. Users should only have permissions necessary for their job functions, and regular audits ensure appropriate access. The process involves transferring ownership of resources and using CLI methods for role changes, particularly from Admin or Author to Reader.

The ability to downgrade user roles in Amazon Quick enhances security and cost efficiency by ensuring that users only have the permissions necessary for their job functions. This development signals to builders and PMs the importance of implementing strict access controls, while investors should recognize it as a move towards more robust governance and risk management in cloud services.

Amazon Bedrock AgentCore enables enterprises to evaluate multi-agent systems for accuracy and explainability, addressing challenges in real-world applications. The platform supports built-in and custom evaluators for assessing performance across various dimensions, ensuring agents provide reliable recommendations and clear reasoning.

Amazon Bedrock AgentCore allows enterprises to evaluate multi-agent systems for accuracy and explainability, which is crucial for builders and PMs focusing on deploying reliable AI solutions. For investors, this development signals a growing emphasis on performance assessment in AI, indicating potential for more robust and trustworthy applications in the market.

Amazon Bedrock AgentCore introduces temporal policies to enhance AI agent security by enforcing stateful authorization rules based on an agent's action history, preventing issues like incorrect data handling and unauthorized actions. These policies operate at the gateway level, ensuring agents cannot bypass them, thus maintaining integrity and compliance in AI workflows.

The introduction of temporal policies in Amazon Bedrock AgentCore enhances AI agent security by enforcing stateful authorization rules, which is crucial for builders and PMs focused on compliance and data integrity. For investors, this development signals a stronger commitment to secure AI solutions, potentially reducing risks associated with data mishandling and unauthorized actions.

AWS announces rate limiting for the Amazon Bedrock AgentCore gateway, enabling fine-grained control over AI traffic management. Users can define OAuth or IAM-based rules for requests, connections, and token throughput, ensuring service availability during traffic spikes. This feature supports various target types, including MCP servers and inference models, enhancing performance and security for different user groups.

AWS's introduction of rate limiting for the Amazon Bedrock AgentCore gateway allows builders and PMs to manage AI traffic more effectively, ensuring consistent service availability during high demand. This development is crucial for investors as it enhances the reliability and scalability of AI applications, potentially leading to better user experiences and increased adoption.

Amazon Bedrock AgentCore introduces enhanced security controls for AI agents, enabling temporal policies and rate limiting to manage agent behaviors and costs effectively. With the new Dogwood policy language, organizations can ensure compliance by evaluating sequences of actions, addressing the trust and security challenges that hinder AI adoption.

The introduction of Amazon Bedrock AgentCore's enhanced security controls and the Dogwood policy language allows builders and PMs to implement more sophisticated compliance measures for AI agents, reducing risks associated with trust and security. This development signals to investors that Amazon is addressing critical barriers to AI adoption, potentially increasing the market's confidence in deploying AI solutions safely.

Amazon Bedrock enables compliance with data residency by processing Claude Code in a single AWS Region, specifically London (eu-west-2). The classic Invoke API and IAM conditions facilitate this, while the Mantle endpoint offers simpler routing in other supported Regions. Organizations must choose the appropriate path based on their specific compliance needs and regional availability.

Amazon Bedrock's introduction of single-Region Claude Code processing in London enhances data residency compliance, allowing organizations to manage sensitive data within specific geographic boundaries. This development is crucial for builders and PMs focused on regulatory adherence and offers investors insights into the growing demand for compliant AI solutions in various sectors.

Amazon Bedrock introduces Agent Skills to streamline the Automated Reasoning policy lifecycle, enabling coding agents to build, test, and validate policies effectively. The suite consists of six skills that enhance the agent's capabilities, ensuring compliance with formal logic and improving the accuracy of AI responses.

Amazon Bedrock's introduction of Agent Skills for Automated Reasoning enhances the policy lifecycle management for AI applications, allowing builders and PMs to implement more accurate and compliant AI solutions. This development signals a shift towards more robust AI governance, which is crucial for investors focused on risk management and regulatory compliance in AI deployments.

PDI Technologies developed PDI Brew, enabling non-technical employees to create web applications on AWS without developer involvement, leveraging Amazon Bedrock for AI capabilities. This agentic app deployer streamlines internal tool delivery, removing traditional bottlenecks in deployment pipelines.

The development of PDI Brew, an agentic app deployer using Amazon Bedrock and AWS Lambda, allows non-technical employees to create web applications independently, significantly reducing reliance on developers. This innovation streamlines internal tool delivery, which can enhance productivity and speed up project timelines for builders, PMs, and investors looking to optimize resource allocation and operational efficiency.

Amazon SageMaker Python SDK v3 introduces generative AI inference recommendations, automating deployment optimization through benchmarking and data-driven configurations. Users can now streamline their workflow by generating ranked deployment options and deploying them directly from notebooks.

The introduction of generative AI inference recommendations in Amazon SageMaker Python SDK v3 allows builders and PMs to automate deployment optimization, significantly reducing the time and effort required for model deployment. For investors, this signals a competitive edge in AI infrastructure, as streamlined workflows can lead to faster product iterations and improved ROI.

LendingTree developed a multi-agent mortgage assistant using Amazon Bedrock, featuring a supervisor and two specialized agents for education and matching, ensuring compliance and user data protection. This AI-driven solution enhances the home-buying experience by providing tailored mortgage options and clear guidance.

LendingTree's development of a multi-agent mortgage assistant using Amazon Bedrock demonstrates how AI can streamline complex processes like home-buying by providing tailored solutions while ensuring compliance. This signals to builders and PMs the potential for creating specialized AI systems that enhance user experience and operational efficiency in financial services.

Mobileye leveraged Amazon Bedrock AgentCore to deploy an AI Support Agent, reducing response times by 90% and achieving over 95% accuracy, while eliminating infrastructure management. This transformation allowed skilled engineers to focus on complex issues instead of routine inquiries, enhancing operational efficiency across the company.

Mobileye's implementation of Amazon Bedrock AgentCore to deploy an AI Support Agent demonstrates a significant advancement in operational efficiency, reducing response times by 90% and achieving over 95% accuracy. For builders and PMs, this signals the potential of AI to streamline support operations, allowing teams to focus on more complex challenges, while investors can recognize the value in AI-driven cost savings and productivity enhancements.

AWS has developed an MCP bridge that enables its cloud-hosted AI agent to access local tools like Excel, enhancing productivity for financial analysts. This solution utilizes WebSocket and native messaging to connect remote clients with local MCP servers, facilitating seamless communication and tool invocation.

AWS's development of the MCP bridge allows cloud-hosted AI agents to interact with local tools like Excel, significantly enhancing productivity for financial analysts. This integration signals a shift towards more versatile AI applications that can seamlessly operate across different environments, presenting opportunities for builders to create more efficient workflows and for investors to back solutions that improve operational efficiency in various industries.

Amazon Bedrock AgentCore now integrates with n8n, enabling the creation of production-ready AI agents with persistent memory and tool access. Users can build agents using various models like OpenAI and Google Gemini without writing infrastructure code, streamlining workflow automation.

The integration of Amazon Bedrock AgentCore with n8n allows builders and PMs to create production-ready AI agents without needing extensive infrastructure coding, significantly reducing development time and complexity. For investors, this signals a growing trend towards accessible AI solutions that can enhance workflow automation across various industries, potentially leading to increased market opportunities.

Amazon Bedrock now includes a built-in Web Search feature, allowing foundation models to access up-to-date web knowledge seamlessly, enhancing accuracy and reducing hallucinations without third-party dependencies. This tool simplifies integration for developers and ensures compliance by keeping data within AWS environments.

The introduction of the Web Search feature in Amazon Bedrock allows foundation models to access real-time web data, improving accuracy and reducing hallucinations. This development simplifies integration for builders and PMs by keeping data within AWS, which enhances compliance and reduces reliance on third-party services, making it a more attractive option for investors in AI infrastructure.

Formula 1® partnered with AWS to develop the Data Accelerator, reducing data source onboarding from 6-8 weeks to approximately 40 minutes using agentic AI on Amazon Bedrock AgentCore. This transformation enhances data integrity and operational efficiency, enabling real-time fan engagement across multiple platforms.

Formula 1's partnership with AWS to create the Data Accelerator, which reduces data onboarding from weeks to minutes, highlights the potential of agentic AI in streamlining operations. This development signals to builders and PMs the importance of integrating advanced AI solutions for enhanced efficiency, while investors should note the competitive advantage gained through rapid data processing capabilities.

Amazon Quick introduces the Agentic Catalog Experience, enabling seamless integration of upstream metadata for AI-driven analytics. This workflow allows data curators to leverage natural language for asset discovery and automatically create datasets with inherited semantics, reducing time to insights from weeks to hours.

The introduction of the Agentic Catalog Experience in Amazon Quick allows builders and PMs to streamline data integration and analytics workflows, significantly reducing the time required to derive insights. For investors, this development signals a competitive advantage in the AI analytics space, potentially increasing market share for Amazon's cloud services.

Amazon Quick introduces the Agentic Catalog Experience, enhancing AI-driven analytics by integrating upstream metadata from data catalogs like AWS Glue and Databricks. This workflow enables data curators to efficiently discover, create, and inherit semantic context, significantly reducing the time to actionable insights from weeks to hours.

The introduction of the Agentic Catalog Experience in Amazon Quick allows builders and PMs to streamline data analytics workflows by drastically reducing the time needed to derive insights from weeks to hours. This development signals a shift towards more efficient data management practices, making it a critical consideration for investors looking to support data-driven companies.