What happened
Amazon Web Services published its weekly announcements roundup on August 10, and several items matter beyond the developer crowd. The headline change is Web Search on Amazon Bedrock: OpenAI models hosted on Bedrock, including the GPT-5.4, GPT-5.5, and GPT-5.6 families, can now browse and retrieve current information from the internet instead of being limited to their training data, and AWS says queries stay within its infrastructure to preserve data residency. AWS Lambda, the pay-per-use compute service behind many business applications, now supports up to 3,000 megabits per second of network bandwidth per function, up from a 625 Mbps starting point, scaling with allocated memory. DynamoDB, AWS's managed database, gained real-time vector search, which powers AI features like semantic search without bolting on a separate database. AWS also open-sourced Dogwood, a governance language for setting rules on what AI agents may do, and announced support for Agent Plugins, an open packaging standard so agent extensions work across tools like Kiro, VS Code, and Cursor.
Why it matters for your business
If your company's software runs on AWS — or your developer or agency builds there — these updates lower the cost and complexity of adding AI features that use current, real-world information rather than stale training data. The data-residency angle matters for DMV-area firms with compliance obligations, including government contractors, because it keeps AI lookups inside AWS rather than routing data to a third party. The Lambda and DynamoDB changes are quieter wins: faster data processing and simpler architectures usually translate to lower cloud bills. If you have an AWS-based product, it is worth a short conversation with your development team about whether any of this replaces something you currently pay a separate vendor for.