Skip to content
Nov 14Sat
  1. AWS Machine Learning Blog42

    Amazon Bedrock AgentCore runtime V2 version optimizes session startup speed

    The Amazon Bedrock AgentCore runtime V2 version significantly improves session startup speed and consistency through technologies such as pre-prepared environments, snapshot recovery, and memory recycling. The V2 version also supports build hooks to load and initialize resources before session startup. Through optimization, AgentCore runtime can handle sessions more efficiently, reduce latency, and improve resource utilization.

Nov 13Fri
  1. AWS Machine Learning Blog42

    AWS Agent Toolkit for AWS Amazon Bedrock migration guide for knowledge base management

    AWS releases a migration guide for Amazon Bedrock knowledge base management, helping users migrate knowledge bases to Bedrock through the Agent Toolkit for AWS, enhancing content management and retrieval capabilities. The migration process includes configuration mapping, content import, application updates, and verification, aiming to simplify operations and improve retrieval accuracy.

Nov 12Thu
Nov 11Wed
  1. AWS Machine Learning Blog62

    AWS Launches Universal Agent Evaluation Framework (UAEF)

    AWS has launched the Universal AI Agent Evaluation Framework (UAEF), which is a framework-agnostic evaluation system for evaluating AI agents. UAEF can convert tracking records from different frameworks into a universal format and use 36 built-in metrics to score 7 dimensions.

    Why it matters: The original text provides changes in capabilities and open entry points, allowing readers to judge how it will change the existing workflow.

  2. AWS Machine Learning Blog60

    Amazon Bedrock Releases Observability Data Lake Function

    Amazon Bedrock has released the observability data lake function, allowing users to centrally view and analyze model call logs, including prompts, model responses, token counts, model usage, etc., to optimize workflows and costs.

    Why it matters: The original article provides changes in capabilities and open entry points, allowing readers to judge how it will change the existing workflow.

  3. Google DeepMind85

    DeepMind releases Gemini 4 Carbon model

    DeepMind has released the new Gemini 4 Carbon model, which boasts cutting-edge performance in agent workflows for cybersecurity capabilities, complex knowledge work, software development, and scientific reasoning. The Carbon model is available for use via API on Google AI Studio and Gemini for business, and offers entry-level pricing for paying developers and enterprises.

    Why it matters: The original article introduces the enhanced capabilities and open entry of the Gemini 4 Carbon model, allowing readers to judge how it will change existing workflows.

Nov 7Sat