AWS used its August 10, 2026 Weekly Roundup to report the AWS Heroes Summit and several AI-focused launches from the previous week, including Web Search on Amazon Bedrock, Runtime Instances on Amazon Bedrock AgentCore, vector search for Amazon DynamoDB, AWS Transform continuous modernization, Dogwood, Agent Plugins, and Kiro Crew.

The AWS Heroes Summit was an invite-only annual gathering for AWS Heroes from around the world. AWS said the summit brought global experts in AI, serverless, and containers into direct collaboration, technical deep-dives, and feedback sessions with internal AWS product and service teams. Day 1 included a fireside chat from AWS CEO Matt Garman, and Day 2 included an AMA with James Hamilton and product-team breakout sessions.

Amazon Bedrock added Web Search for OpenAI models GPT-5.4, GPT-5.5, and GPT-5.6 Sol/Terra/Luna, letting AI applications browse and retrieve internet information beyond their training data while maintaining data residency inside a secured AWS environment with zero data egress. Amazon Bedrock AgentCore added Runtime Instances so developers can deploy and run AI agents on dedicated execution environments with more control, predictable performance, and predictable cost.

Amazon DynamoDB added vector search, allowing developers to store and query vector embeddings (numeric representations of meaning) beside existing DynamoDB data without managing a separate vector database. AWS said the feature supports semantic retrieval over memory used by AI agents for agentic grounding. AWS Transform continuous modernization also became generally available to help engineering teams analyze and remediate technical debt across source code repositories at scale, including mainframe and legacy workloads.

AWS Lambda functions gained higher network bandwidth for data-intensive workloads, scaling from 625 Mbps at 2 GB of memory to up to 3,000 Mbps at 10 GB for functions outside a VPC with 2 GB of memory or more. AWS also open-sourced Dogwood, a governance language for AI agents that supports Cedar policies and temporal conditions, and AgentCore introduced temporal policies based on an agent session’s action history.

AWS said it now supports Agent Plugins, an open source, vendor-neutral specification for packaging AI agent extensions once and shipping them to clients such as Kiro, VS Code, Cursor, or any tool implementing the specification. Kiro Crew was introduced as a persistent, self-evolving workspace for collaborative multi-agent development workflows in the Kiro IDE, spanning repositories, tools, and days.

AWS used its August 10, 2026 Weekly Roundup to report the AWS Heroes Summit and several AI-focused launches from the previous week, including Web Search on Amazon Bedrock, Runtime Instances on Amazon Bedrock AgentCore, vector search for Amazon DynamoDB, AWS Transform continuous modernization, Dogwood, Agent Plugins, and Kiro Crew.

The AWS Heroes Summit was an invite-only annual gathering for AWS Heroes from around the world. AWS said the summit brought global experts in AI, serverless, and containers into direct collaboration, technical deep-dives, and feedback sessions with internal AWS product and service teams. Day 1 included a fireside chat from AWS CEO Matt Garman, and Day 2 included an AMA with James Hamilton and product-team breakout sessions.

Amazon Bedrock added Web Search for OpenAI models GPT-5.4, GPT-5.5, and GPT-5.6 Sol/Terra/Luna, letting AI applications browse and retrieve internet information beyond their training data while maintaining data residency inside a secured AWS environment with zero data egress. Amazon Bedrock AgentCore added Runtime Instances so developers can deploy and run AI agents on dedicated execution environments with more control, predictable performance, and predictable cost.

Amazon DynamoDB added vector search, allowing developers to store and query vector embeddings (numeric representations of meaning) beside existing DynamoDB data without managing a separate vector database. AWS said the feature supports semantic retrieval over memory used by AI agents for agentic grounding. AWS Transform continuous modernization also became generally available to help engineering teams analyze and remediate technical debt across source code repositories at scale, including mainframe and legacy workloads.

AWS Lambda functions gained higher network bandwidth for data-intensive workloads, scaling from 625 Mbps at 2 GB of memory to up to 3,000 Mbps at 10 GB for functions outside a VPC with 2 GB of memory or more. AWS also open-sourced Dogwood, a governance language for AI agents that supports Cedar policies and temporal conditions, and AgentCore introduced temporal policies based on an agent session’s action history.

AWS said it now supports Agent Plugins, an open source, vendor-neutral specification for packaging AI agent extensions once and shipping them to clients such as Kiro, VS Code, Cursor, or any tool implementing the specification. Kiro Crew was introduced as a persistent, self-evolving workspace for collaborative multi-agent development workflows in the Kiro IDE, spanning repositories, tools, and days.