From portal-hopping to instant answers: HEMA’s journey with MCP and Amazon Bedrock

This week, we examine how established organizations can leverage sophisticated AI to consolidate internal knowledge, turning fragmented information into immediate, actionable insights for their teams. The AWS Machine Learning team shares a case study about HEMA, a century-old Dutch retailer, which developed an internal AI assistant named HAL using Amazon Bedrock AgentCore and Model Context Protocol (MCP). This system delivers governed organizational knowledge directly within existing tools, eliminating the need for teams to navigate multiple portals or possess AWS credentials, all while maintaining robust security through Microsoft Entra ID. Essentially, it's about building a secure, intelligent layer over your company's dispersed data, making institutional knowledge instantly accessible and tightly controlled. This approach profoundly affects any organization struggling with information silos or internal knowledge sprawl, offering a pathway to significant efficiency gains. Consider a mid-sized e-commerce operation in Chicago like "Prairie Commerce," where customer service reps currently sift through product databases, shipping FAQs, and internal policy documents across several platforms. Implementing a similar AI assistant could allow them to ask a single natural language question like "What's the return policy for a damaged item shipped to California?" and receive an immediate, accurate, and policy-compliant answer, drastically reducing resolution times and training overhead. For an indie SaaS founder based in Austin running a project management tool, the same principle applies to developer documentation and internal support knowledge; a securely integrated AI could help new hires onboard faster and existing engineers find critical API details without interrupting senior colleagues. Even a small law firm in Boston could benefit by securely centralizing complex case precedents and internal best practices, enabling paralegals to quickly retrieve relevant information without extensive manual searches, improving research efficiency and accuracy. The core opportunity lies in turning your internal knowledge base, no matter how distributed, into a unified, intelligent resource accessible via tools your teams already use daily. This frees up countless hours otherwise spent searching, asking, or waiting for answers, allowing staff to focus on higher-value tasks. The security and governance aspects of MCP are particularly noteworthy, ensuring that sensitive information remains protected and access is appropriately managed without complicating the user experience. To put this idea into practice, consider one specific internal knowledge gap within your organization that frequently consumes team time. Identify the disparate sources of information relevant to that gap. Then, explore how a secure, intelligent agent could unify these sources and present answers in a user-friendly format, perhaps starting with a simple proof-of-concept using a small, well-defined dataset.