Three reference patterns—catalog-first, direct source, and hybrid—show how AI agents can query enterprise systems where data lives using MCP servers and Amazon Bedrock AgentCore, reducing reliance on data engineers for one-off questions.
Use Model Context Protocol (MCP) servers to present diverse systems behind a uniform interface so agents can discover and query tools without knowing source-specific APIs or query languages.
Three reference patterns—catalog-first, direct source, and hybrid—fit different enterprise needs for batch, streaming, and relational data access.
Amazon Bedrock AgentCore Gateway can aggregate MCP servers and handle discovery, authentication, and routing while agents orchestrate queries across batch (S3/Athena), streaming (Kinesis), and relational (Aurora) systems.
# Problem: data silos and the access gap Enterprise data commonly lives in specialized systems that require different expertise: SQL for databases, Athena or Trino for S3 batch data, streaming tooling for Kinesis, and distinct APIs for SaaS applications. That concentrates access knowledge with data engineers and forces business users to file tickets or rely on stale dashboards when they need ad-hoc answers.
Rather than moving or pre-integrating all data, the blog proposes letting AI agents query each system directly through a standardized protocol. The Model Context Protocol (MCP) wraps diverse data sources behind a uniform interface for tool discovery, invocation, and response handling. Agents use MCP servers to find which tool to call, invoke it, and handle results without needing to know per-source authentication, API, or query language details.
# How the reference architecture fits together The reference architecture in the post targets a streaming media use case but generalizes to mixed sources. Key components:
Request flow example (user question to answer):
# Three reference access patterns
# Demo and deployability The blog's demo uses synthetic data and provides a GitHub repository with source code so readers can deploy the reference architecture. Prerequisites include an AWS account with AgentCore access, AWS CLI configured, Python 3.10+, and Docker or Finch.
# Practical implications for teams Teams can reduce one-off engineering requests by enabling agents to reach disparate systems. MCP servers and AgentCore Gateway define a consistent control plane for discovery, authentication, and routing, while agents handle orchestration and result composition. The approach preserves data in place and leverages existing query engines and catalogs.
# Bottom line
<figure data-wp-context="{"imageId":"6aa39718c4dea"}" data-wp-interactive="core/image" data-wp-key="6aa39718c4dea" class="wp-block-image size-large wp-lightbox-container"><img data-recalc-dims="1" fetchpriority="high" decoding="async" width="900" height="506" data-attachment-id="15150" data-permalink="https://digitalth
AI applications used to be relatively simple. A user sent a prompt. An application sent that prompt to a model. The model returned an answer. The application displayed it. That architecture is changing quickly. Modern AI applications increasingly retrieve information, call external APIs, execute tools, interact with da
Large agent prompts often begin as a practical shortcut: policies, domain rules, tool descriptions, examples, recovery procedures, and integration notes are placed in one system message so every capability is always available. That approach stops scaling once an agent accumulates dozens of tools and specialized workflo
When I first started building enterprise applications with Large Language Models (LLMs), I fell into a trap that almost every developer encounters. I thought that scaling an AI system simply meant refining a single, massive prompt. I wrote complex system instructions, packed the context window with rules, and expected
Migrate a multi-model healthcare AI agent from self-managed Amazon ECS with AWS Fargate to Amazon Bedrock AgentCore runtime, preserving triple-model orchestration and vector-enhanced knowledge retrieval while reducing infrastructure management. The framework-agnostic pattern applies across healthcare, financial service
Build a multi-account architecture that keeps each team's data in its own AWS account while giving AI agents a unified way to query across them. A central platform account runs the agent using Amazon Bedrock AgentCore Gateway and MCP, while line-of-business accounts expose their data as MCP servers with secure cross-ac
Loading more related stories...
Open the app view to save this story, compare related coverage, and continue from the same source.
