Amazon Web Services rolled out updates during the week of August 10, 2026, highlighted by native web search capabilities for OpenAI models on Amazon Bedrock, vector search integration for Amazon DynamoDB, and a dedicated AWS Heroes Summit featuring AWS CEO Matt Garman.
Inside the AWS Heroes Summit and Executive Keynotes
Community collaboration took center stage as AWS brought together global experts specializing in AI, serverless, and containers for the invite-only AWS Heroes Summit. Day 1 opened with an inspiring fireside chat delivered by AWS CEO Matt Garman, setting a collaborative tone for technical deep-dives and product feedback sessions.
Day 2 featured an AMA session with James Hamilton, alongside breakout sessions from various product teams that sparked new ideas. Attendees shared insights and feedback across professional networks like LinkedIn, shaping the trajectory of upcoming cloud services.
Web Search and Dedicated Runtime Instances on Amazon Bedrock
Amazon Bedrock expanded its foundational model capabilities this week by introducing native web search integrations for OpenAI models, specifically supporting GPT-5.4, GPT-5.5, and GPT-5.6 Sol/Terra/Luna variants. This functionality allows AI applications to dynamically retrieve up-to-date information from the internet beyond static training datasets.
Crucially for enterprise compliance, this architecture maintains strict data residency within secured AWS environments with zero data egress. Developers building autonomous applications can leverage these tools alongside the official Amazon Bedrock User Guide.
Complementing these retrieval updates, AWS launched Runtime Instances on Amazon Bedrock AgentCore. Engineering teams can now deploy and run AI agents on dedicated runtime instances, gaining more control over execution environments while securing predictable performance and cost.
Vector Search Lands in Amazon DynamoDB
Managing separate vector databases for generative AI workloads just got simpler. Amazon DynamoDB now natively supports vector embeddings, allowing developers to store and query semantic vectors right alongside existing operational data without managing separate vector databases.
DynamoDB already supports storing memory for AI agents. With the addition of native vector search, applications can execute semantic retrieval directly over that memory for agentic grounding—all while maintaining predictable performance associated with DynamoDB.
Modernization at Scale and Network Bandwidth Upgrades
Legacy system remediation moved into general availability with AWS Transform. The continuous modernization tool helps engineering teams analyze and remediate technical debt across source code repositories at scale, enabling automated mainframe and legacy modernization as an ongoing process rather than a one-time migration event. Developers can experiment with these capabilities using the AWS Transform Kiro Power and agent plugins.
At the network layer, AWS Lambda received a throughput boost. Functions configured with 2 GB of memory or more that sit outside a VPC now scale network bandwidth proportionally, from 625 Mbps at 2 GB up to 3,000 Mbps at 10 GB. This upgrade enables data-intensive workloads and faster communication between Lambda functions and other AWS services.
Open Governance and Multi-Agent Workflows with Dogwood and Kiro Crew
AWS open-sourced Dogwood, a purpose-built governance language for AI agents designed to support Cedar policies and add temporal conditions. Powering Dogwood, Amazon Bedrock AgentCore introduced temporal policies that evaluate decisions based on the history of an agent’s actions within a session, rather than the current request alone.
Ecosystem interoperability also expanded with AWS support for Agent Plugins. This open-source, vendor-neutral specification provides a common packaging format, allowing developer extensions to be written once and deployed across multiple clients, including Kiro, VS Code, Cursor, and other tools that implement the spec.
Finally, AWS introduced Kiro Crew, a persistent and self-evolving workspace built into the Kiro IDE. Kiro Crew enables collaborative multi-agent development workflows that span multiple days, repositories, and tools. Developers can run several efforts in parallel or hand work to subagents that report back, removing bottlenecks in complex software engineering pipelines.
The 30-Second Verdict
This week’s updates cement AWS’s focus on enterprise-grade AI autonomy and operational efficiency. By combining native web search and vector memory search inside Bedrock and DynamoDB with open-source governance through Dogwood, AWS is minimizing infrastructure friction for developers building production-ready autonomous agents.
