TLDR: Amazon Web Services (AWS) is making significant strides in artificial intelligence, rolling out new tools under Amazon Bedrock AgentCore to enable secure and scalable deployment of AI agents for enterprises. A key development includes a partnership with Coveo, leveraging its Passage Retrieval API with Amazon Bedrock Agents to automate citation validation and formatting for large language models, thereby enhancing the accuracy and trustworthiness of AI-generated responses.
Amazon Web Services (AWS) is pushing the boundaries of artificial intelligence, unveiling a suite of innovations designed to make AI agents more robust, secure, and accurate for enterprise applications. These advancements, highlighted at the AWS Summit New York 2025 and in recent AWS blog posts, underscore the company’s commitment to transforming how businesses leverage AI.
Central to these developments is Amazon Bedrock AgentCore, a comprehensive set of services aimed at enabling organizations to deploy and operate highly effective AI agents securely and at scale. Announced on July 16, 2025, AgentCore addresses the critical challenge of moving AI agent prototypes to production, offering solutions for runtime, memory, identity, gateway, code interpretation, browser interaction, and observability. Swami Sivasubramanian, AWS VP for Agentic AI, emphasized at the AWS Summit that AI agents are poised to ‘dramatically accelerate innovation and improve productivity across every industry,’ representing a ‘tectonic change’ in software development and interaction.
Among AgentCore’s seven core services, the AgentCore Code Interpreter stands out for its ability to allow AI agents to write and execute code securely in sandbox environments. This capability is crucial for performing complex calculations, processing data, and, significantly, validating reasoning—a foundational element for ensuring the accuracy of AI outputs. Furthermore, the AgentCore Browser Tool provides a secure, cloud-based browser for agents to interact with websites, facilitating tasks like form completion and navigation, which are vital for data gathering and verification.
A particularly impactful development, detailed in an AWS blog post on September 3, 2025, focuses on enhancing the accuracy of Large Language Models (LLMs) through automated citation validation and formatting. AWS has partnered with Coveo, integrating its Passage Retrieval API with Amazon Bedrock Agents. This solution directly tackles the challenge of LLMs generating misleading or inaccurate responses by grounding them in relevant, context-aware enterprise knowledge. The Coveo Passage Retrieval API is designed to pass ranked text passages, or ‘chunks,’ retrieved from a unified index, along with crucial metadata such as source URLs for citations. This ensures that AI-generated answers are not only accurate but also verifiable, instilling greater confidence in their reliability.
This integration means that Amazon Bedrock Agents act as an ‘intelligent backbone,’ interpreting natural language queries and orchestrating the retrieval process to deliver grounded, contextually relevant insights. The Coveo AI-Relevance Platform, with its two-stage retrieval process and machine learning relevancy, extracts the most precise information from structured and unstructured content sources, including critical citation links. This redefines Retrieval Augmented Generation (RAG) systems, moving beyond basic vector search to offer a dynamic and intelligent retrieval pipeline tailored for enterprise needs, complete with native access control enforcement and deep analytics for performance tracking.
AWS is also backing these innovations with substantial investment, announcing an additional $100 million investment in the AWS Generative AI Innovation Center. This funding aims to accelerate the development and deployment of autonomous, agentic AI systems, building on the center’s success in empowering thousands of customers globally. Companies like Warner Bros. Discovery Sports Europe and BMW are already seeing transformative results by leveraging AWS’s agentic AI solutions for tasks ranging from research assistance for commentators to diagnosing network issues in connected vehicles.
Also Read:
- AWS to Unveil Advanced Generative AI and Media Solutions at IBC 2025
- Amazon Unveils ‘Quick Suite’ AI Software to Accelerate Business Decisions, Intensifying AI Agent Competition
These combined efforts from AWS signify a major leap forward in making AI agents not just powerful, but also trustworthy and transparent, addressing critical enterprise needs for accuracy and data integrity in the rapidly evolving AI landscape.
