.jpg)
Amazon Web Services announced significant new capabilities for Amazon Bedrock AgentCore, its framework for building AI agents. The updates focus on production readiness, introducing features for safety, quality evaluation, and enhanced memory to help enterprises build, deploy, and scale agentic AI with greater confidence.
Amazon Bedrock AgentCore launches three new key capabilities: Policy, Evaluations, and episodic Memory.
Policy actively blocks unauthorized agent actions with real-time, deterministic controls.
Evaluations provides a framework for developers to continuously inspect and measure agent behavior quality.
episodic Memory enables agents to learn from past experiences to improve future decision-making.
The enhancements aim to accelerate the path to production for enterprise AI agents.
A wide range of organizations, from Workday to Thomson Reuters, are cited as using AgentCore.
A major new feature is AgentCore Policy, which provides a security layer that operates outside of the agent's own code. It enforces real-time, deterministic controls to actively block unauthorized or unsafe actions that an agent might attempt. This addresses a critical concern for enterprise deployment by allowing safety guardrails to be defined and enforced independently, reducing the risk of agents executing harmful operations.
To help developers maintain and improve agent performance, AgentCore Evaluations offers a framework for continuous inspection. It allows teams to systematically measure the quality of an agent based on its behavior over time. This capability is essential for monitoring drift, validating improvements, and ensuring agents operate as intended as they scale and evolve in production environments.
AgentCore Memory is enhanced with new episodic functionality. This allows agents to learn from and reference past experiences, interactions, and outcomes. By building a memory of events, agents can make more informed and contextually relevant decisions, moving beyond stateless interactions to exhibit more adaptive and intelligent behavior over time.
These innovations are designed to address the core challenges of taking AI agents from proof-of-concept to production. By combining safety (Policy), observability (Evaluations), and advanced cognition (Memory), AWS is providing a more comprehensive toolkit for enterprises. The announcement is underscored by a list of diverse organizations already using AgentCore, indicating growing adoption across industries for operational and customer-facing AI agents.
The new capabilities reinforce AWS's commitment to providing a full-stack environment for agentic AI, focusing not just on the creation of agents but on the governance, evaluation, and continuous learning required for reliable, large-scale deployment.
About AWS
Amazon Web Services (AWS) is guided by customer obsession, pace of innovation, commitment to operational excellence, and long-term thinking. By democratizing technology for nearly two decades and making cloud computing and generative AI accessible to organizations of every size and industry, AWS has built one of the fastest-growing enterprise technology businesses in history. Millions of customers trust AWS to accelerate innovation, transform their businesses, and shape the future. With the most comprehensive AI capabilities and global infrastructure footprint, AWS empowers builders to turn big ideas into reality.
TechIntelPro empowers B2B technology decision-makers with the latest industry trends, deep insights, leadership perspectives and actionable intelligence to drive both organizational and professional milestones—and thrive on the tech pulse.