AWS launches Web Search on Amazon Bedrock AgentCore for grounded AI agents
AWS introduces Web Search on Amazon Bedrock AgentCore, enabling AI agents to ground responses in current, cited web knowledge securely within AWS environments, simplifying integration and infrastructure management.
Today, AWS announced the general availability of Web Search on Amazon Bedrock AgentCore, a fully managed tool enabling AI agents to ground responses in current, cited web knowledge with zero data egress from the customer's secured AWS environment. Web Search uses a built-in connector on the Bedrock AgentCore Gateway via the Model Context Protocol (MCP). Agents send natural-language queries and receive relevant snippets, source URLs, titles, and publication dates to produce grounded responses. Built on Amazon's search infrastructure and informed by experience from Alexa+, Amazon Quick, and Kiro, it combines Amazon's web index with structured knowledge graph data, including the Amazon Knowledge Graph, for verified facts and more accurate results than traditional web search alone. This launch allows developers to build agents without manually adding web search or managing infrastructure, while complying with enterprise governance policies by keeping queries within AWS. To get started, users create the Bedrock AgentCore Gateway with the Web Search tool target in the Bedrock AgentCore console, selecting MCP target protocol and Connectors target type. Interaction is possible via API calls, CLI, or MCP Inspector, an interactive developer tool. Customers such as Benchling and Gen Digital have praised the secure, governed environment and enhanced capabilities. Web Search on Bedrock AgentCore is generally available in the US East (N. Virginia) region with usage-based pricing at $7 per 1,000 queries and free tier credits for new AWS customers. More details and documentation are available on the AWS site.