How DXC Technology built a multi-model AI substrate on Amazon Bedrock
DXC Technology runs mission-critical managed IT services for regulated organizations across 70 countries. Learn how DXC built DCX OASIS, its Human+AI operating model, on a governed multi-model substrate in Amazon Bedrock, separating a resilient model substrate from a trusted harness that orchestrates, governs, and audits AI agents at scale.
DXC Technology, a global IT services provider, collaborated with Amazon to develop a multi-model AI substrate on Amazon Bedrock. This substrate serves as the foundation for DXC OASIS, a system designed to manage AI agents across enterprise IT operations. The primary goal was to create a trusted harness that governs model access at the infrastructure level, ensuring safety and compliance.
Amazon Bedrock provides a unified interface to multiple foundation models, including Anthropic Claude, Amazon Nova, Meta Llama, and open-weight families. This allows DXC OASIS to utilize the most appropriate model for each task, based on factors such as cost, latency, and capability. Amazon Bedrock also offers security and networking controls, keeping all traffic within the AWS perimeter.
The substrate's design prioritizes resilience, flexibility, and regulatory compliance. To achieve this, DXC OASIS employs a supervisor agent and capability agents. The supervisor agent delegates tasks to the capability agents, which draw on a knowledge base, AWS Secrets Manager, Amazon EventBridge, and Amazon S3. All model calls leave the subnet through a VPC interface endpoint and AWS PrivateLink, ensuring that inference traffic does not traverse the public internet.
In summary, DXC Technology built a multi-model AI substrate on Amazon Bedrock to provide a trusted, governed foundation for DXC OASIS. This substrate ensures resilience, flexibility, and regulatory compliance, allowing DXC to deliver measurable business outcomes for its customers while minimizing risk.
Written by urgent.news from AWS Machine Learning's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
This story
This is one outlet's version. Read the fullest account.
