AWS DevOps Agent and Security Agent Reach General Availability
AWS announced general availability of the AWS DevOps Agent and the AWS Security Agent during the week of April 6, 2026, following previews introduced at re:Invent under the "frontier agents" label. Both agents represent a new category in the AWS portfolio: long-running, autonomous systems that pursue operational goals without constant human supervision, distinct from the short-lived Lambda functions or Bedrock agent invocations that previously defined AWS's AI automation surface.
The AWS DevOps Agent targets cloud operations work — investigating incidents, correlating signals across services, reducing time to resolution, and proactively flagging conditions before they escalate. Preview customers report up to 75% lower mean time to resolution (MTTR) and three to five times faster resolution across incident classes. The agent is not limited to AWS infrastructure: it now extends to Azure workloads, correlating data across multicloud deployments to provide unified incident response. T-Mobile, United Airlines, and Western Governors University are cited as early production users. AWS will begin charging for the DevOps Agent starting April 10, 2026.
The AWS Security Agent brings continuous, context-aware penetration testing into the development lifecycle. Rather than scheduling point-in-time pen tests, it operates as a persistent agent that understands the current state of an application's infrastructure and surfaces exploitable configurations and attack paths as they appear — functioning, in AWS's framing, like an on-call human penetration tester. Both agents work across AWS, multicloud, and on-premises environments, positioning them as infrastructure-agnostic operations tools rather than AWS-lock-in plays.
Read more — AWS Cloud Operations Blog
Google Cloud: Gemini 3.1 Flash-Lite, Official BigQuery JDBC/ODBC Drivers
Google Cloud's first week of April 2026 release notes include two updates of particular relevance to Java developers and data platform engineers.
Gemini 3.1 Flash-Lite is now available as the fastest and most cost-efficient model in the Gemini 3 series, positioned for high-volume production workloads where per-token cost matters more than peak reasoning quality. It is available alongside Gemini 3.1 Pro (in preview) on Vertex AI, giving teams a clear cost-to-capability spectrum to optimise against when choosing models for different agent tasks or batch inference pipelines.
Google released an official Google-built JDBC driver for BigQuery in preview, designed for Java applications that need direct, standards-compliant access to BigQuery without relying on the Simba-based third-party drivers that have been the common solution. A companion ODBC driver is also in preview for non-Java application stacks. Both drivers are purpose-built by Google for the BigQuery execution model, providing direct high-performance connections with better support for BigQuery-specific types, streaming inserts, and storage read APIs than generic drivers.
Cloud Composer 3 gained Gemini Cloud Assist investigations for failed Airflow tasks: clicking "Investigate" on a failing task now triggers an AI-assisted root cause analysis that queries logs, task configuration, and dependency graphs before surfacing a diagnosis. This makes operational debugging of complex DAGs accessible without context-switching into log explorers or writing manual diagnostic queries.
Read more — GCP Release Notes
Links & Sources
You May Also Enjoy
