What changed
DXC Technology has built a multi-model AI substrate on Amazon Bedrock, providing a governed gateway to foundation models such as Anthropic Claude, Amazon Nova, Meta Llama, and open-weight families. This substrate allows for secure, flexible, and scalable AI operations across enterprise environments.
What this means in practice
DXC OASIS, the AI harness built on this substrate, enables organizations to route tasks to the most appropriate model based on cost, latency, and capability. The system ensures data remains within AWS boundaries, with customer data not shared with third-party model providers. Amazon Bedrock Guardrails enforce content filtering, PII redaction, and denied-topic policies uniformly across all models. AWS PrivateLink keeps inference traffic within the customer’s VPC, and serverless scaling avoids self-managed GPU infrastructure as a failure point.
Limitations and availability
Model availability varies by AWS Region, and customers must refer to the 'Supported models by AWS Region' section in Amazon Bedrock for details. The DXC OASIS architecture uses Amazon Route 53 and an Application Load Balancer within the customer’s VPC, with the DXC OASIS agent runtime running on Amazon EKS in a private subnet. Inference traffic leaves the subnet through a VPC interface endpoint and AWS PrivateLink to reach Amazon Bedrock, ensuring it does not traverse the public internet. The system also incorporates AWS IAM, AWS KMS, TLS 1.2, Amazon CloudWatch, VPC Flow Logs, AWS CloudTrail, and OpenTelemetry for security and observability.
