# How CrossnoKaye grounds physical AI with Amazon Bedrock AgentCore

- Source: AWS Machine Learning Blog
- Published: 2026-11-09T18:55:43.000Z
- Bad Influence: https://badinfluence.news/items/s1oep7bctlcgp3gs8rzy40ffd
- Original: https://aws.amazon.com/blogs/machine-learning/how-crossnokaye-grounds-physical-ai-with-amazon-bedrock-agentcore/

## Summary

AWS explains how CrossnoKaye built its ATLAS platform and AURA agent system on Amazon Bedrock AgentCore for industrial plant operations. AURA reads plant topology, control logic, and historian data, then runs Python calculations in isolated AgentCore sandboxes that verify outputs before any conclusion is published. The post also describes Chats, Tasks, and Routines, and reports results across more than 100 facilities, including investigations completing in under four minutes.

## Why it matters

The post explains how an agent reads plant control logic and runs verified Python in isolated sandboxes, a reusable pattern for grounding agents in operational systems.