New in Amazon Bedrock AgentCore: Build agents with broader knowledge and continuous learning

2026-06-17 · Source: Artificial Intelligence · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure, Robotics & Autonomous Systems · Depth: Intermediate, long

Summary

Amazon Bedrock AgentCore introduces new capabilities to enhance AI agent development, focusing on broader knowledge access, continuous learning, and stronger controls. Key updates include the generally available Bedrock Managed Knowledge Base, which connects agents to internal data sources like SharePoint and S3 using agentic retrieval, and Web Search on AgentCore, a new tool for real-time web information and Amazon's knowledge graph. For paid content, AgentCore payments (preview) enable agents to access premium services, while WAF AI traffic monetization (GA) allows content owners to charge for access. Optimization features, such as failure, intent, and trajectory insights (preview), help identify silent failures and usage patterns. Recommendations, batch evaluation, and A/B testing (all GA) facilitate data-driven improvements. Additionally, Bedrock Guardrails integration (GA) provides deterministic security controls against prompt injection and harmful content. The AgentCore harness, now generally available, offers a managed orchestration layer, decoupled from models, enabling rapid agent deployment and flexible model switching.

Key takeaway

For MLOps Engineers building or optimizing AI agents, Amazon Bedrock AgentCore's new features streamline deployment and enhance reliability. You should utilize the Managed Knowledge Base and Web Search to provide comprehensive data access, ensuring your agents are grounded in both internal and real-time external information. Employ the optimization capabilities, including insights and A/B testing, to continuously improve agent performance and proactively address silent failures. Integrate Bedrock Guardrails to establish deterministic security controls, protecting your agents from prompt injection and sensitive data exposure in production.

Key insights

AI agents require comprehensive knowledge access, continuous optimization, and robust security controls to reach their full potential in production.

Principles

• Agent performance hinges on context and feedback.
• Deterministic guardrails secure probabilistic agents.
• Decouple agent harness from underlying models.

Method

AgentCore's optimization loop involves understanding agent behavior via production traces, generating data-grounded recommendations, validating fixes with batch evaluation, and A/B testing changes in live traffic.

In practice

• Connect internal data via Managed Knowledge Base.
• Use Web Search for real-time external grounding.
• Implement Guardrails for agent security.

Topics

• Amazon Bedrock AgentCore
• AI Agents
• Knowledge Bases
• Web Search
• Agent Optimization
• AI Security Guardrails
• Managed Harness

Best for: CTO, VP of Engineering/Data, Director of AI/ML, AI Engineer, Machine Learning Engineer, MLOps Engineer

Related on AIssential

• AWS Bedrock Agents: The Beginner’s Guide — AWS Bedrock Agents provide a managed platform for building and deploying AI agents with integrated orchestration and security.
• Powering agentic AI sales strategy with Amazon Bedrock AgentCore — Orchestrating specialized AI agents with a unified interface significantly reduces cognitive load and improves operational efficiency.
• Secure AI agents with Policy in Amazon Bedrock AgentCore — External policy enforcement provides deterministic, auditable security for AI agents, independent of their internal reasoning.
• Semantic search across MCP tools with Amazon Bedrock AgentCore Gateway — Amazon Bedrock AgentCore Gateway enables scalable LLM agents through built-in semantic search for tool selection.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence.