From portal-hopping to instant answers: HEMA’s journey with MCP and Amazon Bedrock
HEMA built an AI assistant, HAL, that pulls all their internal knowledge into one place and delivers it inside the tools people already use, cutting portal‑hopping to seconds.
Why Now
HEMA’s growing engineering teams struggled with fragmented internal knowledge, prompting a move from portal‑hopping to a single AI‑powered knowledge layer built on Amazon Bedrock AgentCore and MCP.
What Happened
HEMA created a service catalog mapping teams, services, APIs, and business capabilities, and built a knowledge layer called HAL using Model Context Protocol (MCP) and Amazon Bedrock AgentCore. MCP standardizes interfaces to internal knowledge sources, while Bedrock AgentCore provides gateway, identity, runtime, memory, and guardrails. The solution now lets engineers ask questions in IDEs or chat windows and get instant answers, replacing the need to navigate multiple portals.
Why It Matters
By centralizing knowledge and delivering it where teams already work, HEMA reduces onboarding time, eliminates inconsistent answers, and frees engineers to focus on coding. The architecture also sets the stage for future action‑layer capabilities, potentially automating procedural tasks.
The Limitation
The article focuses on a single retailer’s internal deployment; results may differ for organizations with different data governance or security requirements.
What You Can Do
Try building a simple MCP tool that exposes a small internal dataset and integrate it with Bedrock AgentCore to see how quickly you can get instant answers in your IDE.
Source
Read original sourceWhy we picked this
Core AI content: building an internal AI assistant on Amazon Bedrock.