Developers demonstrated how to build an agentic AI “semantic layer” for customer-360 analytics on AWS using Amazon Bedrock AgentCore with third-party semantic-layer platforms (AtScale, Strategy Mosaic, and Denodo).
Deployment Provisional 38% confidence first seen
Across three walkthroughs, a Strands agent hosted on Amazon Bedrock AgentCore was connected to different enterprise semantic-layer systems (AtScale, Strategy Mosaic, and Denodo) to answer customer-360 questions. In each setup, the semantic layer planned and executed queries (including joins and metric definitions) and exposed governed access to underlying data sources such as Amazon Aurora and Redshift, with identity-driven controls applied.
Amazon Bedrock AgentCore
Amazon Bedrock AgentCore
AtScale
Denodo Platform
Strands
Strategy Mosaic
AWS
Amazon Aurora
Amazon Aurora
Amazon Bedrock
Amazon Redshift
Amazon Redshift
AtScale C2026.7
AtScale MCP server
MCP server
Stardog
VQL
on-premises PostgreSQL
Decision brief
- What changed
- AWS published three developer walkthroughs showing the same customer-360 agent pattern built on Amazon Bedrock AgentCore with three different third-party semantic-layer platforms: AtScale, Strategy Mosaic, and Denodo. In each example, a Strands agent used the semantic layer to plan and execute governed analytics queries across enterprise data sources such as Amazon Aurora, Amazon Redshift, and, in the Denodo example, on-premises PostgreSQL.
- Why it matters
- For leaders evaluating agentic analytics, these walkthroughs show a concrete integration pattern in which business logic, joins, metric definitions, and access controls stay in the semantic layer rather than being embedded in the agent prompt. That matters because it suggests a way to reuse existing data-governance investments while exposing customer-360 analytics to AI agents on AWS, but only within the semantic-layer products and releases demonstrated in the coverage.
- Evidence
- All three sources are AWS Machine Learning technical walkthroughs describing implementations on Amazon Bedrock AgentCore with AtScale, Strategy Mosaic, and Denodo, respectively. The reporting is consistent across the three articles on the core pattern: the agent delegates query planning/execution to the semantic layer, which enforces joins, metrics, and identity-based access over connected data sources.
- What remains uncertain
- The coverage is vendor-authored and demo-oriented, so it does not verify production performance, deployment complexity, total cost, or customer adoption. It also does not establish how portable this pattern is beyond the cited platform versions and configurations, or how well it handles broader governance, latency, and reliability requirements at scale.
- Monitor next
- Watch for independent customer case studies or AWS/partner announcements showing this Bedrock AgentCore plus semantic-layer pattern running in production with measured security, latency, and data-governance outcomes.
Analytical support, not advice — assumptions and open questions stated above.
Source coverage
Build a semantic layer for agentic AI on AWS with AtScale and Amazon Bedrock AgentCore
Amazon Web Services
Build a semantic layer for agentic AI on AWS with Strategy Mosaic and Amazon Bedrock AgentCore
Amazon Web Services
Build a semantic layer for agentic AI on AWS with Denodo and Amazon Bedrock AgentCore
Amazon Web Services