News
Updates, product notes and practical coverage for AI platforms and operators.
Sanoma Learning has launched Sanna, an Amazon Bedrock-based teacher assistant grounded in its curriculum materials. The product was shaped with educators across seven European countries, including a 1,500-teacher classroom pilot. Its design keeps teachers responsible for lesson choices and future grading decisions while using AI to reduce preparation and administrative work.
AWS has published a nine-stage evaluation framework for production agents built with Amazon Bedrock AgentCore. Its internal sales system exposed fluent but serious data errors that ordinary availability monitoring missed. The framework combines pre-release tests, deployment gates, canary scoring, production sampling, drift detection and feedback so quality is measured continuously rather than at launch alone.
AWS has stopped creation of new batch and recurring AgentCore Insights jobs from 16 October while it redesigns the preview feature. Existing recurring jobs continue through 30 October and historical reports remain accessible. Teams using the failure-analysis service need to preserve evidence, review regional data handling and prepare an interim agent-observability workflow.
AWS has introduced Trust in Depth, a five-layer framework for giving AI agents greater autonomy without routing every action through a person. The approach scopes the job, registers agent identity, enforces temporal tool policies outside model reasoning, reconstructs production traces and uses automated reasoning to prove precisely stated rules across possible execution paths.
AWS has published an AgentCore reference system that turns enterprise documents into grounded PowerPoint presentations in under 90 seconds during internal testing. A supervisor agent chooses generation or summarisation tools, while Bedrock Knowledge Bases, OpenSearch, Claude Sonnet 4.6, Titan embeddings, image generation and tenant-isolated infrastructure handle the workflow.
Tamnoon says its Bedrock-based Tami platform reduced critical cloud exposure by more than 90 per cent within 90 days, with no production issues across millions of workloads. Four specialised agents investigate findings, assign conservative safety ratings, prepare rollback-aware remediation plans and generate reviewed code before short-lived credentials execute approved changes.
AWS has released an eight-agent document-processing reference system that can learn extraction instructions for unfamiliar layouts and reuse them later. Textract, S3 Vectors and Aurora DSQL support classification and validation, while AgentCore Runtime, Gateway, Identity and Observability coordinate credentials, tools and monitoring across identification and training workflows.
Orca Security has detailed a production natural-language query agent built on Amazon Bedrock. Discovery AI converts ordinary questions into filters for a 900-model cloud data graph, reducing input tokens by 93 per cent, overall query cost by 90 per cent and end-to-end latency from about 60 seconds to 15 seconds.
AWS has documented global cross-Region inference for OpenAI GPT-6 Sol and Luna from Singapore, Jakarta, Malaysia, Thailand and Taipei, with GPT-6 Astra available from Singapore. Teams can use Bedrock Converse or OpenAI-compatible APIs, while billing, quota tracking, CloudTrail and CloudWatch remain attached to the source region.
A University of Arizona lab has connected Amazon Bedrock, SageMaker AI and HealthOmics into a research pipeline for rare paediatric disorders. The eight-person team uses foundation models to interpret sequencing results and investigate approved drugs that may affect the same biological pathways, expanding research capacity without claiming to replace laboratory or clinical validation.
Daimler Truck and AWS have detailed a multi-agent SRE prototype that converts monitoring alerts into tested pull requests. Bedrock AgentCore hosts the agents, while duplicate filtering, repository allowlists, sandbox validation, temporal authorisation, rate limits and a cost circuit breaker constrain the workflow before engineers review and merge any proposed fix.
AWS has released a reference architecture that processes public pathogen sequencing data with HealthOmics, harmonises resistance-gene results into Iceberg tables and exposes them through a Bedrock AgentCore assistant. Every answer must cite the gene, detecting tool and confidence value, while guardrails keep the system away from unsupported medical or patient advice.
AWS has demonstrated RareDx, a prototype that coordinates 12 specialist agents to support clinical geneticists from referral intake to a structured report. Bedrock AgentCore, Strands Agents and domain tools handle document extraction, phenotype coding, literature and variant work, while the physician remains responsible for clinical reasoning and the final diagnosis.
AWS has published a practical blueprint for university assistants grounded in catalogues, syllabi, policies and advising material. Bedrock Knowledge Bases supplies retrieval and citations, while metadata filters, evaluation sets and Guardrails limit topics, prompt attacks and personal-data exposure. Institutions can choose managed or self-managed retrieval according to cost, control and data needs.
AWS has proposed the MARS framework for measuring generative AI in aerospace and space programs. It combines mission-adjusted returns, governance-inclusive ownership costs, maturity-weighted forecasts and accepted-output efficiency with risk tiers that determine which Bedrock cost optimisations are permitted, aiming to prevent savings from obscuring quality or compliance exposure.
Qlik has explained how it built Qlik Answers on Amazon Bedrock to serve grounded responses across structured analytics and unstructured documents. The layered architecture supports more than 40,000 customers, regional deployment and specialist agents, while Bedrock Guardrails checks grounding so employees can inspect sources rather than relying on unsupported generated answers.
AWS has published a reference workflow that converts DevOps Agent investigations into narrowly scoped remediation proposals using EventBridge, Lambda Durable Functions and Amazon Bedrock. Read-only actions may run automatically, while infrastructure changes pause for human approval, combining faster incident response with allowlisted tools, checkpoints and an auditable control boundary.
Cornerstone OnDemand says its Orion AI multi-agent system reduced database diagnosis from 45 minutes to 10 and cut median alert noise by 65 per cent. Built by three people using Amazon Bedrock, Strands Agents and domain-specific tools, the system separates live operational queries from conversational memory and gates disruptive actions for human confirmation.
AWS has summarised September additions across Amazon Bedrock, AgentCore and Strands. The release set includes OpenAI-powered managed agents, faster serverless runtime, lower-token agent harnesses, a local decision model, broader frontier-model choice and new governed enterprise connectors that keep knowledge bases synchronised with current business information.
AWS has detailed a dual-layer access-control design for retrieval-augmented generation that combines indexed permissions with live checks against authoritative sources. The approach is designed to keep Amazon Quick and Bedrock answers aligned with current SharePoint, Google Drive and Confluence permissions, reducing the risk of stale access rules exposing sensitive material.
AWS has shown how Amazon Bedrock AgentCore payments lets Incarna agents buy model inference one request at a time through BlockRun and the x402 protocol. Infrastructure-enforced budgets, customer-owned wallets and auditable settlement reduced the integration from an estimated two to three months to three days and about 200 lines of code.
Postman has detailed how Agent Mode uses Amazon Bedrock to serve a global community of 40 million developers. Its production design narrows more than 170 tools to about 15 per task, builds purpose-shaped context, requires approval for state-changing actions and uses regional inference, privacy controls and prompt caching at scale.
AWS has added Z.ai’s GLM 5.3 to Amazon Bedrock for eligible enterprise customers. The managed release supports coding and long-horizon agent workflows, a million-token context window, prompt caching, multiple API styles, selectable reasoning effort and cross-Region inference profiles without requiring customers to operate the underlying model infrastructure.
AWS has opened a public preview of Bedrock Managed Agents, developed with OpenAI for stateful agents running on Bedrock. The 29 September announcement describes durable sessions, reusable skills, MCP tools, per-agent IAM roles and human approval points. The preview is limited to three US regions, and customers still pay for inference and supporting AWS resources.
AWS has added xAI’s Grok 4.7 to Amazon Bedrock, a separate availability step from the model’s earlier xAI launch. The 28 September announcement lists US Geo and Global cross-region inference; AWS also documents a 500,000-token context window and several API routes. The useful question for buyers is how model quality, reasoning effort and token consumption balance within Bedrock.
OmiGroup says its OmiKG research proof-of-concept uses Amazon Bedrock to extract relationships from biomedical literature and analyse patient-specific pathways. A small blinded expert evaluation found more representative mechanisms than two comparison systems, but also highlighted citation-fidelity work. The system is not a clinical product; supervised pilots and commercial planning remain future steps.
The Options Clearing Corporation has launched a security-operations agent built with Amazon Bedrock. It follows an investigative loop against the organisation’s security-event data, gathering evidence and deciding what to examine next. OCC says results are interpretable and auditable, with analysts retained in control. The announcement provides a concrete use case but no independent performance figures.
Nasdaq has launched an agentic environment within Calypso for capital-markets and treasury workflows. A first natural-language assistant can query platform data and documentation; future agents are planned. Nasdaq says the hosted environment uses Amazon Bedrock and allows clients to connect proprietary agents through MCP, with sandboxing and oversight. Availability and governance remain client-specific.
AWS and Morningstar have described an AI assistant inside Direct Advisory Suite that prepares client context and guides multi-step adviser workflows. Bedrock AgentCore provides a session-aware runtime, isolation, networking, memory and telemetry. The case study emphasises entitlements, guardrails and auditability; it does not establish that every adviser or region has the same production access.
AWS has added native Salesforce and Zendesk sources to Amazon Bedrock Managed Knowledge Base. The connectors sync Salesforce knowledge articles and Zendesk articles and community posts, with crawling, metadata extraction and incremental updates handled by the service. They reduce custom ingestion work, but teams still need to validate credentials, permissions, content freshness and answer quality.
AWS has introduced skill-focused evaluation methods across Strands Evals and Amazon Bedrock AgentCore Evaluations. Teams can score whether an agent selected a suitable skill and followed its instructions, while Strands also offers a deterministic invocation check. The tools use recorded trajectories or traces, giving developers more precise evidence than a polished final answer alone.
AWS has made OpenAI’s GPT-6 Sol and GPT-6 Luna generally available on Amazon Bedrock. Sol targets recurring complex work and Luna focused, high-volume tasks; both support up to a million tokens of context. This is a Bedrock hosting announcement distinct from OpenAI’s direct API release. Teams should confirm regional routing, endpoint features, pricing and data controls.
AWS has detailed Grok 4.6 on Amazon Bedrock, adding the standard Bedrock Runtime endpoint and Converse API alongside the OpenAI-compatible Mantle route. The model has a 500,000-token context window and four reasoning settings. Teams should distinguish the September documentation from the earlier August launch, and check endpoint-specific features, residency, pricing and safeguards.
AWS has released the next generation of Bedrock AgentCore Runtime in five regions. It reclaims unused memory during sessions and starts new instances from a prepared snapshot, with AWS reporting roughly two-second P75 cold starts in its test. Developers must select platformVersion V2; production gains and costs need checking against their own agents.
AWS has made Moonshot AI’s Kimi K3 generally available through Amazon Bedrock. The open-weight model combines native vision, a million-token context window and explicit prompt caching, with cross-Region inference across Bedrock regions. This is a new hosting option, not a new Kimi model launch; teams should assess regional routing, cost and workload fit.
Grok 4.6 is now generally available through Amazon Bedrock, according to xAI. Supported AWS-region customers can access the flagship model through Bedrock with a 500,000-token context window, configurable reasoning effort and listed pricing of US$2 per million input tokens and US$6 per million output tokens.
AWS has made Amazon Bedrock Managed Knowledge Base generally available as a managed retrieval service for agents and generative AI applications. It combines native enterprise connectors, multimodal parsing, managed storage, access controls and iterative agentic retrieval without requiring customers to assemble a separate vector-search stack.