Amazon Web Services (AWS) today announced the general availability of Web Search on Amazon Bedrock AgentCore, a significant advancement designed to empower AI agents with the ability to ground their responses in current, cited web knowledge. This fully managed tool integrates seamlessly into Bedrock AgentCore, ensuring that all data egress remains within the customer’s secured AWS environment, addressing critical enterprise requirements for data privacy and governance. The introduction of Web Search marks a pivotal moment in the evolution of generative AI, offering a robust solution to the challenge of providing AI agents with up-to-date, factual information beyond their initial training data.
Introduction to Web Search on Amazon Bedrock AgentCore
The core functionality of Web Search on Bedrock AgentCore is its capacity to retrieve and integrate real-time web information directly into an AI agent’s reasoning process. Utilizing a built-in connector target on the Bedrock AgentCore Gateway, which operates via the Model Context Protocol (MCP), agents can dispatch natural-language queries. In response, Web Search delivers the most relevant snippets, complete with source URLs, titles, and publication dates. This structured and cited information allows the underlying large language models (LLMs) to construct more accurate, relevant, and trustworthy responses, significantly mitigating the risk of "hallucinations" – a common challenge where AI models generate plausible but incorrect information.
This development builds upon Amazon’s extensive experience in search infrastructure, drawing insights and technology from established platforms such as Alexa+, Amazon Quick, and Kiro. These consumer-facing and enterprise search solutions have provided a rich foundation for developing a sophisticated multi-source grounding approach. Web Search on Bedrock AgentCore combines Amazon’s vast web index with proprietary structured knowledge graph data, specifically the Amazon Knowledge Graph. This integration of verified facts with standard web results ensures that agents have access to a broader and more reliable spectrum of information, leading to more precise and contextually rich outputs than traditional web search alone might provide.
Addressing Key Challenges in AI Agent Development
The rapid proliferation of generative AI has ushered in an era of intelligent agents capable of performing complex tasks, from customer service and content generation to data analysis and scientific research. However, a persistent limitation for these agents has been their reliance on static training data, which quickly becomes outdated in a fast-evolving world. Furthermore, the inherent probabilistic nature of LLMs can lead to inaccurate or fabricated responses, eroding user trust and limiting their utility in critical business applications. Web Search on Bedrock AgentCore directly tackles these challenges by providing a dynamic, real-time knowledge base.
The Power of Real-Time, Grounded Information
Traditional AI agents, while powerful, often struggle with questions requiring current events, trending topics, or newly published data. Their responses are constrained by the cutoff date of their training datasets. Web Search on Bedrock AgentCore liberates agents from this constraint, enabling them to fetch the latest facts and developments as needed. This capability is crucial for use cases where timeliness is paramount, such as financial analysis, legal research, news summarization, or competitive intelligence. By grounding responses in fresh, cited web knowledge, agents can deliver actionable insights that reflect the current state of affairs, moving beyond mere summarization of historical data.
Leveraging Amazon’s Extensive Search Infrastructure
The foundation of Web Search’s efficacy lies in Amazon’s decades of experience in building and refining search technologies. The insights gained from powering agentic search experiences across consumer products like Alexa+ (Amazon’s enhanced AI assistant offering a more personalized and proactive experience) and enterprise solutions like Amazon Quick (a search service for enterprise data) and Kiro (a specialized knowledge retrieval system) have been meticulously integrated. This legacy ensures that Web Search on Bedrock AgentCore benefits from a highly optimized, scalable, and resilient search infrastructure. The combination of Amazon’s vast web index, which continuously crawls and indexes billions of web pages, with the curated and verified facts from the Amazon Knowledge Graph, offers a unique advantage. This hybrid approach allows agents to retrieve not just broad web results but also structured, authoritative information, leading to a higher degree of relevance and factual accuracy in their generated responses.
Technical Architecture and Seamless Integration
Integrating Web Search into an AI agent on Bedrock AgentCore has been designed for simplicity and efficiency, allowing developers to focus on agent logic rather than infrastructure management. The service leverages the Model Context Protocol (MCP), an open standard for tool orchestration, which facilitates communication between the agent and various tools, including Web Search.
To get started, developers create a Bedrock AgentCore Gateway, specifying the Web Search tool target. This gateway acts as an intermediary, directing natural language queries from the agent to the Web Search service. The process involves selecting "MCP target" as the protocol and "Connectors" as the target type within the Bedrock AgentCore console. From there, "Web Search tool" can be chosen as a preconfigured target, streamlining the setup. Once the Gateway URL is established, interactions can occur via API calls, Command Line Interface (CLI), or the MCP Inspector, an interactive developer tool designed for testing and debugging MCP servers. The MCP Inspector provides a user-friendly interface to submit queries and observe the results, including snippets, links, and metadata, thereby accelerating the development and testing cycle.
Ensuring Security and Data Governance
A paramount concern for enterprises adopting generative AI is data security and compliance. Web Search on Bedrock AgentCore addresses this by ensuring "zero data egress from customer’s secured AWS environment." This means that user prompts and retrieval queries never leave the trusted AWS infrastructure to interact with external, third-party search API providers. This critical feature allows organizations to maintain strict enterprise governance policies, protecting sensitive data and intellectual property while still leveraging the power of real-time web information. For industries with stringent regulatory requirements, such as healthcare, finance, and government, this secure architecture is not just a feature but a necessity, enabling broader adoption of AI agents in sensitive workflows.
Practical Applications and Customer Success Stories
The immediate impact of Web Search on Bedrock AgentCore is evident across various industries, enhancing the capabilities of AI agents in diverse applications. Early access customers have already begun to leverage this tool to transform their operations.
Benchling: Accelerating Scientific Discovery
Benchling, a leading provider of R&D cloud software for life scientists, has integrated Web Search to empower its AI solutions. Nicholas Larus-Stone, Head of AI Agents at Benchling, highlighted the transformative potential: "Scientists using Benchling AI can now ask about a target they’re actively working on and get answers grounded in both their institutional data in Benchling and published literature. The result is more complete science, and hypothesis generation done right." This capability allows researchers to bridge the gap between proprietary internal data and the vast, ever-growing body of external scientific publications, fostering more robust hypothesis generation and accelerating drug discovery and biological research. The security guarantees of Web Search ensure that sensitive institutional data remains protected while interacting with public scientific knowledge.
Gen Digital: Enhancing Cyber Safety and Online Reputation
Gen Digital, a global leader in consumer and small business cyber safety with brands like Norton, Avast, and LifeLock, is utilizing Web Search to enrich its offerings. Iskander Sanchez-Rola, Senior Director of AI & Innovation at Gen Digital, noted the value for their Norton Revamp service: "With the Web Search tool on Amazon Bedrock AgentCore, Norton Revamp helps professionals build their online reputation with current, grounded content ideas shaped by what’s actually happening in the world today." This application demonstrates how real-time web knowledge can be crucial for services that rely on understanding current trends, public sentiment, and reputational dynamics. Sanchez-Rola emphasized the security aspect, stating, "What we value most is that AWS uses its own search index and keep queries within our trusted AWS environment," reinforcing the importance of data residency and security for enterprise adoption.
These testimonials underscore the versatility and critical value of Web Search on Bedrock AgentCore across different sectors, from highly specialized scientific research to broad consumer cyber safety, all benefiting from enhanced factual accuracy, relevance, and secure data handling.
Availability, Pricing, and Future Outlook
Web Search on Amazon Bedrock AgentCore is now generally available in the US East (N. Virginia) Region. AWS has indicated that regional availability will expand, with details available on the AWS Capabilities by Region page, signifying a planned broader rollout to meet global demand.
The pricing model for Web Search on Bedrock AgentCore is designed for simplicity and scalability, operating on a usage-based structure with no upfront commitments. Customers are charged based on the number of search queries their agents submit to the web search service. The pricing is set at $7 per 1,000 queries. To encourage adoption and allow new customers to explore its capabilities, AWS is offering up to $200 in Free Tier credits, providing a cost-effective entry point for experimentation and development. This transparent and flexible pricing model aligns with AWS’s broader strategy of making advanced AI capabilities accessible to businesses of all sizes.
Strategic Implications for the Enterprise AI Landscape
The launch of Web Search on Bedrock AgentCore represents a strategic move by AWS to further solidify its position in the rapidly evolving generative AI market. By offering a fully managed, secure, and highly integrated web search capability, AWS addresses a critical pain point for enterprises building AI agents: the need for timely, accurate, and verifiable information. This service enhances the overall utility and trustworthiness of AI agents developed on the Bedrock platform, making them more suitable for high-stakes business processes.
This development also intensifies the competition in the AI infrastructure space. As other cloud providers and AI companies race to offer comprehensive platforms for agent development, AWS’s emphasis on deep integration with its secure, enterprise-grade search infrastructure provides a significant differentiator. It enables organizations to accelerate their AI initiatives with confidence, knowing that their agents can access the vastness of the internet without compromising data security or governance. The ability to ground AI responses in current, cited web knowledge is not merely an incremental improvement; it is a foundational enhancement that unlocks new possibilities for AI agents across virtually every industry, from personalized customer experiences and dynamic market analysis to advanced research and development. This tool empowers developers to build a new generation of intelligent agents that are not only conversational but also factually robust and contextually aware, driving innovation and efficiency across the enterprise.
Getting Started with Web Search on Bedrock AgentCore
For developers and organizations eager to leverage this new capability, AWS provides comprehensive documentation and tools. The Bedrock AgentCore Gateway documentation offers detailed guides on configuring and interacting with the Web Search tool. Practical steps include creating a Bedrock AgentCore Gateway, adding the Web Search tool target as a preconfigured connector, and then using invocation code snippets (available for Python with API requests, MCP Python SDK, Strands MCP Client, and MCP Inspector) to integrate web search functionality into AI agents.
Feedback channels are open through AWS re:Post for Amazon Bedrock AgentCore and standard AWS Support contacts, allowing users to share their experiences and contribute to the ongoing refinement of the service. This commitment to continuous improvement, combined with the robust capabilities and security features of Web Search on Amazon Bedrock AgentCore, positions it as a cornerstone for future advancements in enterprise AI. The announcement, initially made on June 17, 2026, and updated on June 18, 2026, with a clear pricing statement, underscores AWS’s dedication to providing transparent and powerful tools for the generative AI ecosystem.
