GUIDE · MODELS · #2260
Strategy Mosaic exposes MCP access for agentic AI with Amazon Bedrock AgentCore (Strategy One, Mar 2026)
A Strategy blog post demonstrates using Strategy Mosaic as a dimensional semantic layer for a customer-360 pattern across Amazon Redshift and Amazon Aurora and exposing certified models via the Mosaic Model Context Protocol (MCP) to a Strands agent hosted on Amazon Bedrock AgentCore. The walkthrough (source of truth: Strategy One, March 2026) covers modeling, metric syntax, linked models with no ETL, per-user RLS/OLS propagation, live vs in-memory execution, and publishing the same governed definitions to BI, notebooks, and custom apps.
KEY POINTS
- A Strategy blog post demonstrates using Strategy Mosaic as a dimensional semantic layer for a customer-360 pattern across Amazon Redshift and Amazon Aurora and exposing certified models via the Mosaic Model Context Protocol (MCP) to a Strands agent hosted on Amazon Bedrock AgentCore.
- The walkthrough (source of truth: Strategy One, March 2026) covers modeling, metric syntax, linked models with no ETL, per-user RLS/OLS propagation, live vs in-memory execution, and publishing the same governed definitions to BI, notebooks, and custom apps.
- By making Mosaic models reachable via MCP for Bedrock-hosted agents, enterprises can serve governed dimensional metrics and security (RLS/OLS) directly to agentic workflows without in-agent logic or ETL.
WHY IT MATTERS
By making Mosaic models reachable via MCP for Bedrock-hosted agents, enterprises can serve governed dimensional metrics and security (RLS/OLS) directly to agentic workflows without in-agent logic or ETL.