AWS Launches AgentCore Optimization in Preview
SEATTLE — Amazon Web Services this week introduced AgentCore Optimization in public preview, adding a performance-tuning capability to its AgentCore platform for building and deploying AI agents.
The new feature introduces what AWS calls an “agent performance loop” — a system designed to iteratively measure and improve how AI agents perform in enterprise environments, according to an AWS blog post published this week.
AgentCore Optimization expands AWS’s enterprise agentic AI offering, which competes with platforms from Microsoft, Google Cloud, and startups targeting the agent infrastructure market. The tool is part of the broader AgentCore platform that provides enterprises with managed infrastructure for deploying AI agents at scale.
The preview launch comes as enterprises move beyond chatbot deployments toward more complex agentic systems that can take actions, use tools, and complete multi-step workflows autonomously.
AWS’s approach focuses on creating a continuous feedback loop that allows enterprises to monitor agent performance, identify failure points, and apply optimizations without requiring a full rebuild of agent architectures.
The preview release positions AWS alongside competitors providing enterprise-grade agent infrastructure. Microsoft has invested in its Copilot platform and Azure AI Agent Service, while Google Cloud offers its Vertex AI Agent Builder.
AgentCore Optimization is available immediately in preview to AWS customers. Pricing and general availability timelines were not disclosed in the initial announcement.
