October 26, 2026

Using AtScale and Amazon Bedrock AgentCore for an Agentic AI Semantic Layer

← AI for Your Work

WebCrimp analysis and practical explanation. AI-assisted; reviewed before publication.

Quick Summary

The supplied source describes an AWS blog walkthrough about using AtScale as a semantic layer for agentic AI. It follows earlier customer 360 examples with Stardog and Denodo, then uses AtScale dimensional modeling with SML, Git, SQL, XMLA, and MCP. The runtime uses Amazon Bedrock AgentCore, Strands, Amazon Bedrock, and Amazon Redshift.

What Happened?

The post explains how to build a customer 360 semantic layer with AtScale. AtScale models dimensions, hierarchies, facts, measures, calculations, perspectives, row security, and autonomous aggregate management. The model is authored in Semantic Modeling Language YAML and can be exposed through SQL, XMLA, and MCP. A Strands agent on Amazon Bedrock AgentCore plans and writes queries through the AtScale MCP server registered on AgentCore Gateway. Identity propagation and access control are handled through AgentCore Identity and AtScale. Deployment uses Amazon EKS, the AtScale Helm chart, and K8s Blueprints. The example uses retail customer 360 data in Amazon Redshift, with Amazon Aurora as the operational source and replication keeping the analytical copy current. The representative question concerns top customers by lifetime spend who have not enrolled in rewards.

Why It Matters

The walkthrough shows how a semantic layer can sit between an AI agent and enterprise data. Joins, metric definitions, and access rules can live in the semantic layer instead of the prompt. The same computed answer can be used by the agent and existing BI tools, which may reduce inconsistency and support governed self-service analytics.

Who Can Use It?

AI/ML · Cloud · Data Analytics · DevOps · Software Development · Management · Business · Finance · Operations · Sales · Supply Chain

How Can I Use This?

If you work in AI/ML, Cloud, Data Analytics, DevOps, Software Development, Management, Business, Finance, Operations, Sales, Supply Chain

  • For AI/ML, prototype an agent that calls the AtScale MCP server through AgentCore Gateway. For Cloud, design the AWS identity and runtime path. For Data Analytics, map source tables to AtScale dimensions and measures. For DevOps, deploy and operate AtScale on Amazon EKS. For Software Development, build the agent and MCP integration. For Management, use governed metrics for reporting decisions. For Business, frame questions around shared definitions. For Finance, use calculated measures for revenue logic. For Operations, use row security and perspectives. For Sales, use customer 360 views. For Supply Chain, use product hierarchies. Human oversight and anonymized data are required.

AI Tip

Put metric logic in calculated measures and version the semantic model in Git. Review changes through pull requests and human oversight before deployment.

AI Trick

Use a governed metric and dimension model so agent queries and BI tools use the same definitions instead of relying on prompts or ad hoc queries.

Example Workflow

  1. Author the semantic model in SML YAML in Git. Deploy AtScale on Amazon EKS with the Helm chart and K8s Blueprints. Register the AtScale MCP server with AgentCore Gateway. Configure identity and access through AgentCore Identity and AtScale. Run the Strands agent on AgentCore. Validate answers with BI tools and human review. Workflow suggestions need verification.

Copy This Prompt

You are an assistant for a governed semantic layer. Use semantic model [semantic_model] and data source [data_source]. The business question is [business_question]. Use only the defined dimensions, hierarchies, measures, and calculations. Return the query approach, assumptions, and caveats. Do not generate definitions outside the model.

Recommended Tools

  • Amazon Bedrock
  • Amazon Bedrock AgentCore
  • Strands
  • Amazon Redshift
  • Amazon Aurora
  • Amazon EKS
  • Helm
  • Semantic Modeling Language
  • YAML
  • Git
  • Model Context Protocol
  • Amazon Redshift Serverless

Check official documentation, availability and pricing before using these suggestions.

Difficulty

Intermediate

Estimated Time Saved

Not measured; validate with your own workflow.

Source

AWS AI
Build a semantic layer for agentic AI on AWS with AtScale and Amazon Bedrock AgentCore
Original publication date: 2026-10-20T15:33:54+00:00