AI news, product launches and company developments.
OpenAI & AnthropicReported on 2026-10-09: Postman and AWS describe how Agent Mode runs for 40 million developers on Amazon Bedrock, focusing on controlling tool sprawl, exposing schema-based reads, and treating context as the real bottleneck.
On 2026-10-07, AWS Machine Learning Blog reported a framework for AI center of excellence leaders to estimate the value of agentic automation across time savings, exception handling, decision quality, and maintenance economics.
AWS states that Anthropic Claude Opus 5.5 and Claude Sonnet 5.5 are available on Amazon Bedrock in the AWS GovCloud (US) Regions.
Confirmed benchmark guidance published on 2026-09-22 by the AWS Machine Learning Blog: concurrency sweeps can help right-size a generative AI endpoint on Amazon SageMaker AI by systematically benchmarking it at increasing load levels.
Amazon SageMaker Inference now offers prefix-aware routing, which sends requests sharing the same prompt prefix to the same instance so the KV cache stays warm. In benchmarks on Llama 3.1 70B, it reduced P50 time-to-first-token by up to 77% and increased KV cache hit rates from about 25% to over 80%.
Reported on 2026-09-04: AWS describes customizing an Amazon Bedrock knowledge base for large and complex documents by combining Amazon Textract text extraction with Amazon Bedrock generative AI. The post covers ingesting and preprocessing PDFs and images, then querying utility bills at scale.
Reported on 2026-08-26, Amazon Bedrock AgentCore Evaluations decouples agent evaluation from the framework used to build an agent. It can score agents that emit OpenTelemetry telemetry, including agents built with LangGraph, LlamaIndex, the OpenAI Agents SDK, Google ADK, the Claude Agent SDK, or Strands Agents.
On 2026-08-19, AWS published a blog post detailing three serverless patterns—task-token callback, direct service integration, and durable functions—for asynchronously invoking Amazon Bedrock AgentCore agents from AWS Step Functions pipelines. These patterns eliminate idle compute costs while the AI agent processes requests.
Tradeshift deployed Amazon Quick with agentic AI capabilities to replace their legacy BI tool. This transition resulted in query response times up to 30 times faster, a 40 percent reduction in total cost of ownership, and transformed embedded analytics into a revenue-generating product.
Amazon Bedrock now offers a managed knowledge base to build enterprise search for agents. It simplifies setup, improves retrieval accuracy, and supports production readiness. The blog provides code examples to help developers implement these features easily.
Amazon has introduced Grok 4.3 on its Bedrock platform, designed for agentic and enterprise AI workloads. The model supports features like chat, configurable reasoning, tool calling, structured output, image input, and multi-turn conversations. This integration simplifies access to advanced AI capabilities for developers and businesses.
Amazon demonstrates building a restaurant telephony AI host using Amazon Bedrock AgentCore and Amazon Nova 2 Sonic. The system handles voice orders from greeting to confirmation with real-time speech processing. It integrates with restaurant backend systems and optimizes caller experience by warming the agent session before answering.
