Tech Meridian
← ENTITY INDEX
COMPANY · ENTITY #197
Related event timeline, sources and context from the news index.
ShipStation Global built Deckhand, an AI-powered IT help desk agent for Slack, using Amazon Bedrock AgentCore and Anthropic’s Claude. Deckhand draws on a consolidated Bedrock-indexed knowledge base, reads error screenshots, and creates/updates Jira Service Management tickets, reportedly resolving 97.5% of incoming channel queries with a 96% positive employee feedback rating.
SpaceXAI announced Grok 4.7 is available on Amazon Bedrock with features aimed at long-horizon agent workloads: a 500K-token context window, tool calling, configurable reasoning-effort levels (low/medium/high/xhigh), and support for Chat Completions/Responses and the Bedrock Converse API. In SpaceXAI's tests of a tool-calling agent on three search-style tasks, the agent answered 67 of 72 runs correctly (18/18 at xhigh; 15/18 at low), illustrating a tunable trade-off between correctness and token use; Bedrock access uses cross-Region inference profiles and requires specific IAM permissions.
Amazon Bedrock now exposes system-defined global cross-Region inference profiles that let callers in Thailand, Malaysia, Singapore, Indonesia, and Taiwan invoke OpenAI GPT-6 Sol and GPT-6 Luna (and GPT-6 Astra from Singapore) via the bedrock-runtime endpoint. The profiles are callable from the source Region by passing a profile ID as the model (supporting the Converse API and OpenAI-compatible /openai/v1 paths), with billing, CloudTrail, CloudWatch, and quota tracking remaining in the source Region; some Regions must be opt-in in your account.
AWS announced September 2026 updates to Amazon Bedrock, Bedrock AgentCore, and the open-source Strands toolkit: Amazon Bedrock Managed Agents (powered by OpenAI) entered public preview; the AgentCore runtime reduced cold-start latency and memory overhead for serverless agents; Strands released an open-source harness that uses ~28% fewer tokens and the Decider 2B local decision model (~115 ms); and OpenAI’s Astra, Sol, and Luna families (including GPT-6 Astra and Astra Ultrafast) became available on Bedrock alongside updated Claude versions (Fable 5.1, Opus 5.5).
Postman describes how its Agent Mode — an AI-native interface for API testing, documentation, discovery, and implementation — is run in production for ~40 million developers using Amazon Bedrock. The post details architectural patterns (controlling tool sprawl, schema-based reads, treating context as the bottleneck), operational choices (geographically scoped cross-Region inference, model-dependent zero-data-retention, multi-tier prompt caching), human approval before state changes, and use of Amazon Bedrock Guardrails to redact PII; it also explains narrowing >170 tools to ~15 via a vector database of tool embeddings to limit the model's tool surface.
Anthropic’s Claude Opus 5.5 and Claude Sonnet 5.5 (alongside Claude Sonnet 5) are available via Amazon Bedrock in AWS GovCloud (US); Sonnet 5 holds FedRAMP Class D and DoD IL4/IL5 authorizations and Opus/Sonnet 5.5 hold FedRAMP Class D certification on Bedrock (users should verify current model certification status). The post outlines using Anthropic’s agentic coding tool Claude Code with Bedrock’s bedrock-runtime and bedrock-mantle endpoints, describing security features (Zero Operator Access), compliance pathways, and integrations for CI/CD and developer tooling.
Amazon implemented a two-stage access-control approach for Amazon Quick and Amazon Bedrock Knowledge Bases: pre-retrieval ACL filtering against indexed attributes, followed by real-time verification with authoritative data sources (e.g., Google Drive APIs using service-account impersonation) to ensure RAG responses respect current permissions.
Qlik has put Qlik Answers — a grounded, enterprise-scale natural-language assistant — into general availability in February 2026 and built it on Amazon Bedrock for chat, embeddings, streaming, and reranking while using Amazon OpenSearch Service for document retrieval. The system uses a layered architecture (routing, specialist agents, conversational analytics, retrieval, and a model-access gateway with Bedrock Guardrails) and reports early adoption metrics (Discovery Agent >100,000 discoveries and widespread agent usage), with customer deployments at Bystronic, Lintech International, and TouchPoint Support Services.
AWS describes a workflow that connects AWS DevOps Agent investigation events to an automated remediation pipeline: EventBridge triggers a Lambda Durable Functions orchestrator that sends findings to Amazon Bedrock, which selects from an allowlisted set of purpose-built Lambda tools to propose and (for read-only actions) execute fixes or (for mutating changes) prepare pre-validated changes awaiting one-click human approval. The design uses Durable Functions checkpoints to pause for approvals and iteratively call Bedrock until remediation completes, aiming to lower mean time to resolution (MTTR).
A six-week structured program paired business professionals with mentors and production-grade tools (Amazon Bedrock AgentCore, Strands Agents SDK, DynamoDB, Node.js/Python Flask) to close the gap between AI knowledge and hands-on capability; participants built WealthWise, a multi-agent financial advisory prototype using Amazon Nova models and won first place. The article lays out the program structure, barriers encountered, and five principles that supported rapid learning and prototype delivery.
Cornerstone OnDemand built a multi-agent system called Orion AI using Amazon Bedrock and the Strands Agents orchestration framework; the company reports database diagnosis time fell from about 45 minutes to 10 minutes (a 78% reduction), alert noise dropped a median 65%, and a three-person team delivered the system in six months. Orion AI coordinates specialized agents for monitoring, diagnostics, lifecycle operations and remediation, integrates with existing tools (e.g., SQL Server and Jira), and emphasizes data-privacy controls and domain-specific agent design.
AWS published guidance for customers on implementing ISO/IEC 42005:2025, explaining how AI system impact assessments fit into enterprise risk management and pointing to tools such as an ISO/IEC 42001:2023 AIMS implementation guide and the Well‑Architected Framework: Responsible AI Lens. The post also notes AWS has pursued ISO-related certifications for several AI services (Amazon Bedrock, Amazon Q Business, Amazon Transcribe, Amazon Textract) and highlights Annex D and E of ISO/IEC 42005 as paths to integrate or adopt standalone AI impact assessment processes.
Z.ai’s GLM 5.3, a 753‑billion‑parameter mixture‑of‑experts model optimized for coding and long‑horizon agentic tasks and reported to show strong cybersecurity performance, is now available to eligible enterprise customers via Amazon Bedrock. The Bedrock integration offers OpenAI‑compatible APIs (Responses and Chat Completions), Bedrock Invoke/Converse, cross‑Region inference, implicit/explicit prompt caching, and service tiers to simplify managed access and reduce latency and cost.
A technical walkthrough demonstrates using LangChain with Amazon Bedrock Managed Knowledge Bases to compare standard single-shot retrieval (Retrieve API) and multi-step agentic retrieval (AgenticRetrieveStream API). The post explains architecture, IAM permissions, boto3 version requirements (agentic_retrieve_stream requires boto3 >= 1.43.32), and trade-offs including cost and trace event inspection.
Amazon describes Amazon Bedrock AgentCore Evaluations, a fully managed capability for assessing multi-agent system performance with built-in and custom evaluators for dimensions like helpfulness, task success, and explainability, and shows how it pairs with Bedrock Guardrails for runtime safeguards. The post includes a reference implementation using Strands Agents SDK, an orchestrator and specialized sub-agents, Bedrock runtimes, and MCP Server to demonstrate evaluation and explainability in a supply‑chain scenario.
AWS demonstrates an "Adjudicated Query" design pattern and reference architecture that combines a generative-AI chat surface (Amazon Quick) with a deterministic, versioned rules engine to run auditable, provably complete compliance sweeps (example: tens of thousands of leases). The architecture separates conversational translation/narration (model) from pass/fail determinations (rules engine), emits a completeness receipt, and shows an end-to-end deployable sample using MCP, Amazon Bedrock, Cognito, API Gateway, Lambda, and Aurora Serverless v2.
Amazon’s blog post describes using Amazon SageMaker AI’s multi-turn reinforcement learning (MTRL) capability to fine-tune a Qwen3.6-27B model for an LLM-powered search agent. The write-up details MTRL features—serverless execution, modular agent-environment interfaces, asynchronous rollouts, built-in algorithms (PPO, CISPO, IS), trajectory observability via MLflow, and evaluation jobs—and shows an enterprise search use case combining BM25 and vector search tools.
Amazon Payments implemented a multi-objective contextual multi-armed bandit (LinUCB) on Amazon SageMaker AI to personalize content across a three-step acquisition funnel. In a seven-week online A/B test they report a high single-digit percentage relative lift in final-funnel conversion for one customer population while another saw no improvement; the team published the AWS architecture, method details, and a code repository for experimentation on synthetic data.
Uniopen (part of Taiwan’s Uni‑President Enterprises Group) adapted Amazon Nova 2 Lite for its retail moderation taxonomy by supervised fine-tuning in Amazon SageMaker AI and additional prompt‑level output optimization, using Nova 2 Pro to generate candidate corrections that humans verify before inclusion in training data. The deployment pipeline uses Amazon S3, DynamoDB, Argo Workflows on EKS, Argo CD, SNS, CloudWatch and Bedrock Guardrails with hard/soft gate checks, fixed test sets, and mandatory human review for ambiguous cases.
AWS published a walkthrough demonstrating a three-agent music production pipeline deployed on Amazon Bedrock AgentCore Runtime Instances. The example colocates agents on a single EC2-backed instance with persistent volumes and an NVIDIA L4 GPU (running ACE-Step for on-instance music generation), uses shared runtimeSessionId for agent collaboration, and packages agents as ECR containers or S3 artifacts to enable multi-day, stateful workflows.
Amazon Bedrock now supports Anthropic Claude Opus 5 and Claude Sonnet 5 with in-region inference on the bedrock-runtime endpoint: both Opus 5 and Sonnet 5 are available in Seoul (ap-northeast-2), and Sonnet 5 is available in Singapore (ap-southeast-1). The release lets customers call the bedrock-runtime endpoint (InvokeModel, Converse, or Anthropic Messages API) so inference requests and data are processed and logged entirely within the specified AWS Region to help meet data residency requirements.
According to OpenAI, GPT-6.1 Sol is now generally available on Amazon Bedrock, offering stronger reasoning for agentic coding, complex document analysis, and multistep professional workflows. OpenAI claims GPT-6.1 Sol matches GPT-6 Astra on DeepSWE v1.1 at roughly one-fifth the cost per task and outperforms GPT-6 Sol; the model is accessible via Codex, ChatGPT Work and Bedrock APIs and can be run with AWS controls (IAM, CloudTrail, VPC endpoints) and hardware-isolated inference with customer data-not-used-for-training options.
At Dev Day, OpenAI announced updates to its software engineering agent Codex, introducing reusable cloud development environments accessible across devices, a refreshed Codex CLI (including voice-driven task control), a code review integration in the ChatGPT desktop app, and a new Codex Security Cloud for scanning and fixing repo issues. The company also released API updates including a Decisions API powered by Luna and enhancements to the Agents API with support for Amazon Bedrock Managed Agents on AWS.
An AWS how-to describes an architecture for a React web app that converts portfolios of contract PDFs into structured, queryable data: AI extraction agents running via Amazon Bedrock (AgentCore) use Claude Sonnet to extract fields and Claude Haiku to verify them, Amazon Textract acts as a deterministic tiebreaker for signatures, results are stored in Amazon Aurora PostgreSQL, and users query via embedded analytics and a natural-language chat interface with real-time pipeline status over WebSocket.
Condé Nast partnered with the AWS Generative AI Innovation Center to build a multimodal video discovery system using TwelveLabs' Marengo embeddings via Amazon Bedrock and indexed in Amazon OpenSearch Service. The decoupled architecture — embedding generation on Bedrock and low-latency k-NN search on OpenSearch — enables intent-based, image-driven, and timestamped searches across transcripts, audio, and visuals, reducing average discovery time from ~250 minutes to under 2 minutes per task.
xAI has made Grok 4.7 available via Amazon Bedrock (served through the bedrock-runtime endpoint with cross-Region inference profiles). The model offers a 500K-token context window, four configurable reasoning-effort levels (low/medium/high/xhigh), improved long-horizon self-verification for coding and knowledge work, new safeguards, and support for Responses, Chat Completions, InvokeModel and Converse APIs.
Anthropic’s Claude Sonnet 5.5, a faster and more cost‑efficient Sonnet model for focused coding and knowledge‑work tasks, is available on Amazon Bedrock (via the Global CRIS inference profile) and on Claude Platform on AWS in North America. The integration keeps data within AWS infrastructure and works with AWS controls such as IAM, CloudTrail, CloudWatch, and Bedrock Guardrails, and can be called programmatically via the Anthropic Messages API through bedrock-runtime.
NarrateAI’s second engineering post describes five coordinated techniques—adaptive pipeline orchestration, cross-account multi-model failover, real-time streaming evaluation, a composite evaluation framework, and data accuracy verification—implemented on Amazon Bedrock AgentCore to deliver validated real-time conversational responses for enterprise BI; the post reports these techniques achieve approximately 99% numerical accuracy while streaming results. The write-up targets engineers building LLM applications and explains how the layered pipeline maintains throughput, avoids throttling, and catches numerical hallucinations in executive-facing workflows.
Amazon’s post describes a reference architecture for building multi-account AI agents that query data in-place across multiple AWS line-of-business (LOB) accounts without centralizing datasets. It shows how a central platform account can host agents and Bedrock inference while AgentCore Gateway (with MCP, AgentCore Identity, Policy in AgentCore/Cedar, and optional Okta auth) aggregates LOB MCP servers, provides unified tool discovery, authorization, observability, and applies Bedrock guardrails and RAG-backed knowledge retrieval.
Aderant built an Intelligent Ticket Analyzer that uses Amazon Nova Lite through Amazon Bedrock to automate context gathering, classification, routing, and knowledge enrichment for support tickets across 268 client environments. In the first 2.5 weeks (June 30–July 17, 2026) it reviewed 109 tickets with ~96% routing accuracy, estimates saving 8–14 engineering hours per week, and ran at under $30/month operating cost.
Allow Google Analytics to measure visits? It stays off until you choose. You can change your choice at any time.