AWS DevOps Agent hits GA with Datadog, PagerDuty, and LaunchDarkly integrations
AWS made its DevOps Agent generally available, marking a major push into agentic SRE/operations. The agent works with the Datadog MCP Server to autonomously correlate observability data with deployed AWS infrastructure, automating root-cause analysis so production incidents can be resolved 'in minutes rather than hours.' A coordinated wave of integration launches accompanied GA.
The PagerDuty integration helps SRE teams move from 'something broke' alerts to understanding why, cutting the manual log-grepping and deployment-tracing that slow 2 a.m. incident response. The LaunchDarkly integration connects feature flags to incident response — during an outage the agent can identify relevant flags and coordinate disabling them, removing manual latency at the worst possible moment. Kiro-powered capabilities surface metrics, traces, topology, and configurations directly inside the IDE so engineers don't juggle browser tabs when alarms fire. AWS also added Release Management so 'agents don't just write code, they help you ship safely and reliably.'
This positions AWS squarely against a crowded autonomous-ops field including Datadog's own Bits AI, PagerDuty's agentic features, and a wave of incident-response startups. The MCP-centric design — leaning on the Model Context Protocol as the integration substrate — reflects how quickly MCP has become the connective tissue for enterprise agents across vendors.
The skeptical question for SRE leaders is trust and blast radius: autonomous remediation that disables feature flags or pushes config changes during an incident must have tight guardrails and audit trails, or it risks turning one outage into two. AWS's framing emphasizes 'validate' steps, but real-world adoption will hinge on how much autonomy teams are willing to grant in production. Watch for early customer postmortems on whether the agent measurably reduces MTTR.
